├── hardware
    ├── plate.cb
    ├── plate.dxf
    ├── plate.gcode
    └── plate.svg
├── readme.markdown
└── src
    ├── ai
        ├── camera_capture.sh
        ├── camera_watch.sh
        ├── explore_generate_data.ipynb
        ├── explore_run.ipynb
        ├── explore_train.ipynb
        ├── generate_data.py
        ├── run.py
        ├── train.py
        └── video.py
    └── nodebot
        ├── nodebot.js
        ├── package.json
        ├── tm1640_led_screen.js
        ├── tm1640_test.js
        └── udp_receive.js


/hardware/plate.cb:
--------------------------------------------------------------------------------
  1 | <?xml version="1.0" encoding="utf-8"?>
  2 | <CADFile xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema" Version="0.9.8.0" Name="plate">
  3 |   <layers>
  4 |     <layer name="0" color="127,255,0">
  5 |       <ModificationCount>0</ModificationCount>
  6 |       <objects />
  7 |     </layer>
  8 |     <layer name="Base" color="127,255,0">
  9 |       <ModificationCount>0</ModificationCount>
 10 |       <objects>
 11 |         <pline id="32" Closed="true">
 12 |           <ModificationCount>3</ModificationCount>
 13 |           <mat m="Identity" />
 14 |           <pts>
 15 |             <p>20,109.899891,0</p>
 16 |             <p>20,20,0</p>
 17 |             <p>134.899721,20,0</p>
 18 |             <p>134.899721,109.899891,0</p>
 19 |           </pts>
 20 |         </pline>
 21 |         <pline id="33" Closed="true">
 22 |           <ModificationCount>3</ModificationCount>
 23 |           <mat m="Identity" />
 24 |           <pts>
 25 |             <p b="-0.19894233606149078">33.570998,31.57118,0</p>
 26 |             <p b="-0.41419080310176931">32.999861,32.950246,0</p>
 27 |             <p b="-0.41410714293940259">34.95006,34.899888,0</p>
 28 |             <p b="-0.41419129667519039">36.899709,32.950246,0</p>
 29 |             <p b="-0.19893943734096858">34.95006,31.000034,0</p>
 30 |           </pts>
 31 |         </pline>
 32 |         <pline id="34" Closed="true">
 33 |           <ModificationCount>3</ModificationCount>
 34 |           <mat m="Identity" />
 35 |           <pts>
 36 |             <p b="-0.1989398647751052">116.328645,98.329249,0</p>
 37 |             <p b="-0.19894019620934827">116.899788,96.950192,0</p>
 38 |             <p b="-0.19893782725251211">116.328646,95.571133,0</p>
 39 |             <p b="-0.4141909321934">114.949587,94.999991,0</p>
 40 |             <p b="-0.4141900804004896">112.999939,96.950192,0</p>
 41 |             <p b="-0.19893870940794967">114.949587,98.90039,0</p>
 42 |           </pts>
 43 |         </pline>
 44 |         <pline id="35" Closed="true">
 45 |           <ModificationCount>3</ModificationCount>
 46 |           <mat m="Identity" />
 47 |           <pts>
 48 |             <p b="-0.19893930126048295">116.32865,31.57118,0</p>
 49 |             <p b="-0.41419242106117021">114.949587,31.000034,0</p>
 50 |             <p b="-0.41410724632076384">112.999939,32.950246,0</p>
 51 |             <p b="-0.4141898081878661">114.949587,34.899888,0</p>
 52 |             <p b="-0.19893725041748042">116.899788,32.950246,0</p>
 53 |           </pts>
 54 |         </pline>
 55 |         <pline id="36" Closed="true">
 56 |           <ModificationCount>3</ModificationCount>
 57 |           <mat m="Identity" />
 58 |           <pts>
 59 |             <p b="-0.19893872884216024">33.571003,98.329249,0</p>
 60 |             <p b="-0.41419085709084441">34.95006,98.90039,0</p>
 61 |             <p b="-0.41419105061597944">36.899709,96.950192,0</p>
 62 |             <p b="-0.1989375372373709">34.95006,94.999991,0</p>
 63 |             <p b="-0.19893868993716615">33.571002,95.571133,0</p>
 64 |             <p b="-0.19893740402468851">32.999861,96.950192,0</p>
 65 |           </pts>
 66 |         </pline>
 67 |       </objects>
 68 |     </layer>
 69 |   </layers>
 70 |   <MachiningOptions>
 71 |     <FastPlungeHeight>-1</FastPlungeHeight>
 72 |     <Stock>
 73 |       <PMin>0,0</PMin>
 74 |       <PMax>0,0</PMax>
 75 |       <Color>255,165,0</Color>
 76 |     </Stock>
 77 |     <ToolProfile>Unspecified</ToolProfile>
 78 |     <MachiningOrigin />
 79 |   </MachiningOptions>
 80 |   <parts>
 81 |     <part Name="Part1">
 82 |       <machineops>
 83 |         <engrave>
 84 |           <TargetDepth state="Default">-0.4</TargetDepth>
 85 |           <DepthIncrement state="Default">0.4</DepthIncrement>
 86 |           <FinalDepthIncrement state="Default">0</FinalDepthIncrement>
 87 |           <CutOrdering state="Default">DepthFirst</CutOrdering>
 88 |           <StockSurface state="Default">0</StockSurface>
 89 |           <RoughingClearance state="Default">0</RoughingClearance>
 90 |           <ClearancePlane state="Default">3</ClearancePlane>
 91 |           <SpindleDirection state="Default">CW</SpindleDirection>
 92 |           <SpindleSpeed state="Value">60</SpindleSpeed>
 93 |           <SpindleRange state="Default">0</SpindleRange>
 94 |           <VelocityMode state="Default">Undefined</VelocityMode>
 95 |           <primitive>
 96 |             <prim>36</prim>
 97 |             <prim>33</prim>
 98 |             <prim>35</prim>
 99 |             <prim>34</prim>
100 |           </primitive>
101 |           <WorkPlane state="Default">XY</WorkPlane>
102 |           <OptimisationMode state="Default">Experimental</OptimisationMode>
103 |           <RoughingFinishing state="Default">Roughing</RoughingFinishing>
104 |           <ToolDiameter state="Default">0</ToolDiameter>
105 |           <ToolNumber state="Default">0</ToolNumber>
106 |           <ToolProfile state="Default">Unspecified</ToolProfile>
107 |           <PlungeFeedrate state="Default">300</PlungeFeedrate>
108 |           <CutFeedrate state="Value">300</CutFeedrate>
109 |           <MaxCrossoverDistance state="Default">0.7</MaxCrossoverDistance>
110 |           <StartPoint state="Default" />
111 |           <CustomMOPHeader state="Default" />
112 |           <CustomMOPFooter state="Default" />
113 |           <Name>Holes</Name>
114 |         </engrave>
115 |         <engrave>
116 |           <TargetDepth state="Value">-0.4</TargetDepth>
117 |           <DepthIncrement state="Value">0.2</DepthIncrement>
118 |           <FinalDepthIncrement state="Default">0</FinalDepthIncrement>
119 |           <CutOrdering state="Default">DepthFirst</CutOrdering>
120 |           <StockSurface state="Default">0</StockSurface>
121 |           <RoughingClearance state="Default">0</RoughingClearance>
122 |           <ClearancePlane state="Default">3</ClearancePlane>
123 |           <SpindleDirection state="Default">CW</SpindleDirection>
124 |           <SpindleSpeed state="Value">50</SpindleSpeed>
125 |           <SpindleRange state="Default">0</SpindleRange>
126 |           <VelocityMode state="Default">Undefined</VelocityMode>
127 |           <primitive>
128 |             <prim>32</prim>
129 |           </primitive>
130 |           <WorkPlane state="Default">XY</WorkPlane>
131 |           <OptimisationMode state="Default">Experimental</OptimisationMode>
132 |           <RoughingFinishing state="Default">Roughing</RoughingFinishing>
133 |           <ToolDiameter state="Default">0</ToolDiameter>
134 |           <ToolNumber state="Default">0</ToolNumber>
135 |           <ToolProfile state="Default">Unspecified</ToolProfile>
136 |           <PlungeFeedrate state="Default">300</PlungeFeedrate>
137 |           <CutFeedrate state="Value">600</CutFeedrate>
138 |           <MaxCrossoverDistance state="Default">0.7</MaxCrossoverDistance>
139 |           <StartPoint state="Default" />
140 |           <CustomMOPHeader state="Default" />
141 |           <CustomMOPFooter state="Default" />
142 |           <Name>Outside</Name>
143 |         </engrave>
144 |       </machineops>
145 |       <Stock>
146 |         <PMin>0,0</PMin>
147 |         <PMax>0,0</PMax>
148 |         <Color>255,165,0</Color>
149 |       </Stock>
150 |       <ToolProfile>Unspecified</ToolProfile>
151 |       <MachiningOrigin />
152 |       <Nesting>
153 |         <BasePoint />
154 |         <NestMethod>None</NestMethod>
155 |         <Spacing>0</Spacing>
156 |       </Nesting>
157 |       <NestingItem />
158 |     </part>
159 |   </parts>
160 |   <ToolpathVisiblility>All</ToolpathVisiblility>
161 |   <ActiveLayer>Base</ActiveLayer>
162 |   <ActivePart>Part1</ActivePart>
163 | </CADFile>


--------------------------------------------------------------------------------
/hardware/plate.dxf:
--------------------------------------------------------------------------------
   1 |   0
   2 | SECTION
   3 |   2
   4 | HEADER
   5 |   9
   6 | $ACADVER
   7 |   1
   8 | AC1014
   9 |   9
  10 | $HANDSEED
  11 |   5
  12 | FFFF
  13 |   9
  14 | $MEASUREMENT
  15 |  70
  16 |      1
  17 |   0
  18 | ENDSEC
  19 |   0
  20 | SECTION
  21 |   2
  22 | TABLES
  23 |   0
  24 | TABLE
  25 |   2
  26 | VPORT
  27 |   5
  28 | 8
  29 | 330
  30 | 0
  31 | 100
  32 | AcDbSymbolTable
  33 |  70
  34 |      4
  35 |   0
  36 | VPORT
  37 |   5
  38 | 2E
  39 | 330
  40 | 8
  41 | 100
  42 | AcDbSymbolTableRecord
  43 | 100
  44 | AcDbViewportTableRecord
  45 |   2
  46 | *ACTIVE
  47 |  70
  48 |      0
  49 |  10
  50 | 0.0
  51 |  20
  52 | 0.0
  53 |  11
  54 | 1.0
  55 |  21
  56 | 1.0
  57 |  12
  58 | 210.0
  59 |  22
  60 | 148.5
  61 |  13
  62 | 0.0
  63 |  23
  64 | 0.0
  65 |  14
  66 | 10.0
  67 |  24
  68 | 10.0
  69 |  15
  70 | 10.0
  71 |  25
  72 | 10.0
  73 |  16
  74 | 0.0
  75 |  26
  76 | 0.0
  77 |  36
  78 | 1.0
  79 |  17
  80 | 0.0
  81 |  27
  82 | 0.0
  83 |  37
  84 | 0.0
  85 |  40
  86 | 341.0
  87 |  41
  88 | 1.24
  89 |  42
  90 | 50.0
  91 |  43
  92 | 0.0
  93 |  44
  94 | 0.0
  95 |  50
  96 | 0.0
  97 |  51
  98 | 0.0
  99 |  71
 100 |      0
 101 |  72
 102 |    100
 103 |  73
 104 |      1
 105 |  74
 106 |      3
 107 |  75
 108 |      0
 109 |  76
 110 |      0
 111 |  77
 112 |      0
 113 |  78
 114 |      0
 115 |   0
 116 | ENDTAB
 117 |   0
 118 | TABLE
 119 |   2
 120 | LTYPE
 121 |   5
 122 | 5
 123 | 330
 124 | 0
 125 | 100
 126 | AcDbSymbolTable
 127 |  70
 128 |      1
 129 |   0
 130 | LTYPE
 131 |   5
 132 | 14
 133 | 330
 134 | 5
 135 | 100
 136 | AcDbSymbolTableRecord
 137 | 100
 138 | AcDbLinetypeTableRecord
 139 |   2
 140 | BYBLOCK
 141 |  70
 142 |      0
 143 |   3
 144 | 
 145 |  72
 146 |     65
 147 |  73
 148 |      0
 149 |  40
 150 | 0.0
 151 |   0
 152 | LTYPE
 153 |   5
 154 | 15
 155 | 330
 156 | 5
 157 | 100
 158 | AcDbSymbolTableRecord
 159 | 100
 160 | AcDbLinetypeTableRecord
 161 |   2
 162 | BYLAYER
 163 |  70
 164 |      0
 165 |   3
 166 | 
 167 |  72
 168 |     65
 169 |  73
 170 |      0
 171 |  40
 172 | 0.0
 173 |   0
 174 | LTYPE
 175 |   5
 176 | 16
 177 | 330
 178 | 5
 179 | 100
 180 | AcDbSymbolTableRecord
 181 | 100
 182 | AcDbLinetypeTableRecord
 183 |   2
 184 | CONTINUOUS
 185 |  70
 186 |      0
 187 |   3
 188 | Solid line
 189 |  72
 190 |     65
 191 |  73
 192 |      0
 193 |  40
 194 | 0.0
 195 |   0
 196 | ENDTAB
 197 |   0
 198 | TABLE
 199 |   2
 200 | LAYER
 201 |   5
 202 | 2
 203 | 100
 204 | AcDbSymbolTable
 205 |  70
 206 | 2
 207 |   0
 208 | LAYER
 209 |   5
 210 | 50
 211 | 100
 212 | AcDbSymbolTableRecord
 213 | 100
 214 | AcDbLayerTableRecord
 215 |   2
 216 | 0
 217 |  70
 218 | 0
 219 |   6
 220 | CONTINUOUS
 221 |   0
 222 | LAYER
 223 |   5
 224 | 51
 225 | 100
 226 | AcDbSymbolTableRecord
 227 | 100
 228 | AcDbLayerTableRecord
 229 |   2
 230 | Base
 231 |  70
 232 | 0
 233 |   6
 234 | CONTINUOUS
 235 |   0
 236 | ENDTAB
 237 |   0
 238 | TABLE
 239 |   2
 240 | STYLE
 241 |   5
 242 | 3
 243 | 330
 244 | 0
 245 | 100
 246 | AcDbSymbolTable
 247 |  70
 248 |      1
 249 |   0
 250 | STYLE
 251 |   5
 252 | 11
 253 | 330
 254 | 3
 255 | 100
 256 | AcDbSymbolTableRecord
 257 | 100
 258 | AcDbTextStyleTableRecord
 259 |   2
 260 | STANDARD
 261 |  70
 262 |      0
 263 |  40
 264 | 0.0
 265 |  41
 266 | 1.0
 267 |  50
 268 | 0.0
 269 |  71
 270 |      0
 271 |  42
 272 | 2.5
 273 |   3
 274 | txt
 275 |   4
 276 | 
 277 |   0
 278 | ENDTAB
 279 |   0
 280 | TABLE
 281 |   2
 282 | VIEW
 283 |   5
 284 | 6
 285 | 330
 286 | 0
 287 | 100
 288 | AcDbSymbolTable
 289 |  70
 290 |      0
 291 |   0
 292 | ENDTAB
 293 |   0
 294 | TABLE
 295 |   2
 296 | UCS
 297 |   5
 298 | 7
 299 | 330
 300 | 0
 301 | 100
 302 | AcDbSymbolTable
 303 |  70
 304 |      0
 305 |   0
 306 | ENDTAB
 307 |   0
 308 | TABLE
 309 |   2
 310 | APPID
 311 |   5
 312 | 9
 313 | 330
 314 | 0
 315 | 100
 316 | AcDbSymbolTable
 317 |  70
 318 |      2
 319 |   0
 320 | APPID
 321 |   5
 322 | 12
 323 | 330
 324 | 9
 325 | 100
 326 | AcDbSymbolTableRecord
 327 | 100
 328 | AcDbRegAppTableRecord
 329 |   2
 330 | ACAD
 331 |  70
 332 |      0
 333 |   0
 334 | ENDTAB
 335 |   0
 336 | TABLE
 337 |   2
 338 | DIMSTYLE
 339 |   5
 340 | A
 341 | 330
 342 | 0
 343 | 100
 344 | AcDbSymbolTable
 345 |  70
 346 |      1
 347 |   0
 348 | DIMSTYLE
 349 | 105
 350 | 27
 351 | 330
 352 | A
 353 | 100
 354 | AcDbSymbolTableRecord
 355 | 100
 356 | AcDbDimStyleTableRecord
 357 |   2
 358 | ISO-25
 359 |  70
 360 |      0
 361 |   3
 362 | 
 363 |   4
 364 | 
 365 |   5
 366 | 
 367 |   6
 368 | 
 369 |   7
 370 | 
 371 |  40
 372 | 1.0
 373 |  41
 374 | 2.5
 375 |  42
 376 | 0.625
 377 |  43
 378 | 3.75
 379 |  44
 380 | 1.25
 381 |  45
 382 | 0.0
 383 |  46
 384 | 0.0
 385 |  47
 386 | 0.0
 387 |  48
 388 | 0.0
 389 | 140
 390 | 2.5
 391 | 141
 392 | 2.5
 393 | 142
 394 | 0.0
 395 | 143
 396 | 0.03937007874016
 397 | 144
 398 | 1.0
 399 | 145
 400 | 0.0
 401 | 146
 402 | 1.0
 403 | 147
 404 | 0.625
 405 |  71
 406 |      0
 407 |  72
 408 |      0
 409 |  73
 410 |      0
 411 |  74
 412 |      0
 413 |  75
 414 |      0
 415 |  76
 416 |      0
 417 |  77
 418 |      1
 419 |  78
 420 |      8
 421 | 170
 422 |      0
 423 | 171
 424 |      3
 425 | 172
 426 |      1
 427 | 173
 428 |      0
 429 | 174
 430 |      0
 431 | 175
 432 |      0
 433 | 176
 434 |      0
 435 | 177
 436 |      0
 437 | 178
 438 |      0
 439 | 270
 440 |      2
 441 | 271
 442 |      2
 443 | 272
 444 |      2
 445 | 273
 446 |      2
 447 | 274
 448 |      3
 449 | 340
 450 | 11
 451 | 275
 452 |      0
 453 | 280
 454 |      0
 455 | 281
 456 |      0
 457 | 282
 458 |      0
 459 | 283
 460 |      0
 461 | 284
 462 |      8
 463 | 285
 464 |      0
 465 | 286
 466 |      0
 467 | 287
 468 |      3
 469 | 288
 470 |      0
 471 |   0
 472 | ENDTAB
 473 |   0
 474 | TABLE
 475 |   2
 476 | BLOCK_RECORD
 477 |   5
 478 | 1
 479 | 330
 480 | 0
 481 | 100
 482 | AcDbSymbolTable
 483 |  70
 484 |      1
 485 |   0
 486 | BLOCK_RECORD
 487 |   5
 488 | 1F
 489 | 330
 490 | 1
 491 | 100
 492 | AcDbSymbolTableRecord
 493 | 100
 494 | AcDbBlockTableRecord
 495 |   2
 496 | *MODEL_SPACE
 497 |   0
 498 | BLOCK_RECORD
 499 |   5
 500 | 1B
 501 | 330
 502 | 1
 503 | 100
 504 | AcDbSymbolTableRecord
 505 | 100
 506 | AcDbBlockTableRecord
 507 |   2
 508 | *PAPER_SPACE
 509 |   0
 510 | ENDTAB
 511 |   0
 512 | ENDSEC
 513 |   0
 514 | SECTION
 515 |   2
 516 | BLOCKS
 517 |   0
 518 | BLOCK
 519 |   5
 520 | 20
 521 | 330
 522 | 1F
 523 | 100
 524 | AcDbEntity
 525 |   8
 526 | 0
 527 | 100
 528 | AcDbBlockBegin
 529 |   2
 530 | *MODEL_SPACE
 531 |  70
 532 |      0
 533 |  10
 534 | 0.0
 535 |  20
 536 | 0.0
 537 |  30
 538 | 0.0
 539 |   3
 540 | *MODEL_SPACE
 541 |   1
 542 | 
 543 |   0
 544 | ENDBLK
 545 |   5
 546 | 21
 547 | 330
 548 | 1F
 549 | 100
 550 | AcDbEntity
 551 |   8
 552 | 0
 553 | 100
 554 | AcDbBlockEnd
 555 |   0
 556 | BLOCK
 557 |   5
 558 | 1C
 559 | 330
 560 | 1B
 561 | 100
 562 | AcDbEntity
 563 |  67
 564 |      1
 565 |   8
 566 | 0
 567 | 100
 568 | AcDbBlockBegin
 569 |   2
 570 | *PAPER_SPACE
 571 |   1
 572 | 
 573 |   0
 574 | ENDBLK
 575 |   5
 576 | 1D
 577 | 330
 578 | 1B
 579 | 100
 580 | AcDbEntity
 581 |  67
 582 |      1
 583 |   8
 584 | 0
 585 | 100
 586 | AcDbBlockEnd
 587 |   0
 588 | ENDSEC
 589 |   0
 590 | SECTION
 591 |   2
 592 | ENTITIES
 593 |   0
 594 | LINE
 595 |   5
 596 | 100
 597 | 100
 598 | AcDbEntity
 599 |   8
 600 | Base
 601 |  62
 602 | 7
 603 | 100
 604 | AcDbLine
 605 |  10
 606 | 0.050161
 607 |  20
 608 | 89.949761
 609 |  30
 610 | 0.0
 611 |  11
 612 | 0.050161
 613 |  21
 614 | 0.049870
 615 |  31
 616 | 0.0
 617 |   0
 618 | LINE
 619 |   5
 620 | 101
 621 | 100
 622 | AcDbEntity
 623 |   8
 624 | Base
 625 |  62
 626 | 7
 627 | 100
 628 | AcDbLine
 629 |  10
 630 | 0.050161
 631 |  20
 632 | 0.049870
 633 |  30
 634 | 0.0
 635 |  11
 636 | 114.949882
 637 |  21
 638 | 0.049870
 639 |  31
 640 | 0.0
 641 |   0
 642 | LINE
 643 |   5
 644 | 102
 645 | 100
 646 | AcDbEntity
 647 |   8
 648 | Base
 649 |  62
 650 | 7
 651 | 100
 652 | AcDbLine
 653 |  10
 654 | 114.949882
 655 |  20
 656 | 0.049870
 657 |  30
 658 | 0.0
 659 |  11
 660 | 114.949882
 661 |  21
 662 | 89.949761
 663 |  31
 664 | 0.0
 665 |   0
 666 | LINE
 667 |   5
 668 | 103
 669 | 100
 670 | AcDbEntity
 671 |   8
 672 | Base
 673 |  62
 674 | 7
 675 | 100
 676 | AcDbLine
 677 |  10
 678 | 114.949882
 679 |  20
 680 | 89.949761
 681 |  30
 682 | 0.0
 683 |  11
 684 | 0.050161
 685 |  21
 686 | 89.949761
 687 |  31
 688 | 0.0
 689 |   0
 690 | LINE
 691 |   5
 692 | 104
 693 | 100
 694 | AcDbEntity
 695 |   8
 696 | Base
 697 |  62
 698 | 7
 699 | 100
 700 | AcDbLine
 701 |  10
 702 | 0.050161
 703 |  20
 704 | 89.949761
 705 |  30
 706 | 0.0
 707 |  11
 708 | 0.050161
 709 |  21
 710 | 89.949761
 711 |  31
 712 | 0.0
 713 |   0
 714 | SPLINE
 715 |   5
 716 | 105
 717 | 100
 718 | AcDbEntity
 719 |   8
 720 | Base
 721 |  62
 722 | 7
 723 | 100
 724 | AcDbSpline
 725 |  70
 726 | 8
 727 |  71
 728 | 3
 729 |  72
 730 | 8
 731 |  73
 732 | 4
 733 |  74
 734 | 0
 735 |  40
 736 | 0
 737 |  40
 738 | 0
 739 |  40
 740 | 0
 741 |  40
 742 | 0
 743 |  40
 744 | 1
 745 |  40
 746 | 1
 747 |  40
 748 | 1
 749 |  40
 750 | 1
 751 |  10
 752 | 15.000221
 753 |  20
 754 | 78.950260
 755 |  30
 756 | 0.0
 757 |  10
 758 | 16.077117
 759 |  20
 760 | 78.950066
 761 |  30
 762 | 0.0
 763 |  10
 764 | 16.949979
 765 |  20
 766 | 78.076957
 767 |  30
 768 | 0.0
 769 |  10
 770 | 16.949870
 771 |  20
 772 | 77.000062
 773 |  30
 774 | 0.0
 775 |   0
 776 | SPLINE
 777 |   5
 778 | 106
 779 | 100
 780 | AcDbEntity
 781 |   8
 782 | Base
 783 |  62
 784 | 7
 785 | 100
 786 | AcDbSpline
 787 |  70
 788 | 8
 789 |  71
 790 | 3
 791 |  72
 792 | 8
 793 |  73
 794 | 4
 795 |  74
 796 | 0
 797 |  40
 798 | 0
 799 |  40
 800 | 0
 801 |  40
 802 | 0
 803 |  40
 804 | 0
 805 |  40
 806 | 1
 807 |  40
 808 | 1
 809 |  40
 810 | 1
 811 |  40
 812 | 1
 813 |  10
 814 | 16.949870
 815 |  20
 816 | 77.000062
 817 |  30
 818 | 0.0
 819 |  10
 820 | 16.949981
 821 |  20
 822 | 75.923165
 823 |  30
 824 | 0.0
 825 |  10
 826 | 16.077118
 827 |  20
 828 | 75.050055
 829 |  30
 830 | 0.0
 831 |  10
 832 | 15.000221
 833 |  20
 834 | 75.049861
 835 |  30
 836 | 0.0
 837 |   0
 838 | SPLINE
 839 |   5
 840 | 107
 841 | 100
 842 | AcDbEntity
 843 |   8
 844 | Base
 845 |  62
 846 | 7
 847 | 100
 848 | AcDbSpline
 849 |  70
 850 | 8
 851 |  71
 852 | 3
 853 |  72
 854 | 8
 855 |  73
 856 | 4
 857 |  74
 858 | 0
 859 |  40
 860 | 0
 861 |  40
 862 | 0
 863 |  40
 864 | 0
 865 |  40
 866 | 0
 867 |  40
 868 | 1
 869 |  40
 870 | 1
 871 |  40
 872 | 1
 873 |  40
 874 | 1
 875 |  10
 876 | 15.000221
 877 |  20
 878 | 75.049861
 879 |  30
 880 | 0.0
 881 |  10
 882 | 14.482980
 883 |  20
 884 | 75.049808
 885 |  30
 886 | 0.0
 887 |  10
 888 | 13.986908
 889 |  20
 890 | 75.255258
 891 |  30
 892 | 0.0
 893 |  10
 894 | 13.621163
 895 |  20
 896 | 75.621003
 897 |  30
 898 | 0.0
 899 |   0
 900 | SPLINE
 901 |   5
 902 | 108
 903 | 100
 904 | AcDbEntity
 905 |   8
 906 | Base
 907 |  62
 908 | 7
 909 | 100
 910 | AcDbSpline
 911 |  70
 912 | 8
 913 |  71
 914 | 3
 915 |  72
 916 | 8
 917 |  73
 918 | 4
 919 |  74
 920 | 0
 921 |  40
 922 | 0
 923 |  40
 924 | 0
 925 |  40
 926 | 0
 927 |  40
 928 | 0
 929 |  40
 930 | 1
 931 |  40
 932 | 1
 933 |  40
 934 | 1
 935 |  40
 936 | 1
 937 |  10
 938 | 13.621163
 939 |  20
 940 | 75.621003
 941 |  30
 942 | 0.0
 943 |  10
 944 | 13.255418
 945 |  20
 946 | 75.986748
 947 |  30
 948 | 0.0
 949 |  10
 950 | 13.049968
 951 |  20
 952 | 76.482820
 953 |  30
 954 | 0.0
 955 |  10
 956 | 13.050022
 957 |  20
 958 | 77.000062
 959 |  30
 960 | 0.0
 961 |   0
 962 | SPLINE
 963 |   5
 964 | 109
 965 | 100
 966 | AcDbEntity
 967 |   8
 968 | Base
 969 |  62
 970 | 7
 971 | 100
 972 | AcDbSpline
 973 |  70
 974 | 8
 975 |  71
 976 | 3
 977 |  72
 978 | 8
 979 |  73
 980 | 4
 981 |  74
 982 | 0
 983 |  40
 984 | 0
 985 |  40
 986 | 0
 987 |  40
 988 | 0
 989 |  40
 990 | 0
 991 |  40
 992 | 1
 993 |  40
 994 | 1
 995 |  40
 996 | 1
 997 |  40
 998 | 1
 999 |  10
1000 | 13.050022
1001 |  20
1002 | 77.000062
1003 |  30
1004 | 0.0
1005 |  10
1006 | 13.049969
1007 |  20
1008 | 77.517303
1009 |  30
1010 | 0.0
1011 |  10
1012 | 13.255419
1013 |  20
1014 | 78.013374
1015 |  30
1016 | 0.0
1017 |  10
1018 | 13.621164
1019 |  20
1020 | 78.379119
1021 |  30
1022 | 0.0
1023 |   0
1024 | SPLINE
1025 |   5
1026 | 10a
1027 | 100
1028 | AcDbEntity
1029 |   8
1030 | Base
1031 |  62
1032 | 7
1033 | 100
1034 | AcDbSpline
1035 |  70
1036 | 8
1037 |  71
1038 | 3
1039 |  72
1040 | 8
1041 |  73
1042 | 4
1043 |  74
1044 | 0
1045 |  40
1046 | 0
1047 |  40
1048 | 0
1049 |  40
1050 | 0
1051 |  40
1052 | 0
1053 |  40
1054 | 1
1055 |  40
1056 | 1
1057 |  40
1058 | 1
1059 |  40
1060 | 1
1061 |  10
1062 | 13.621164
1063 |  20
1064 | 78.379119
1065 |  30
1066 | 0.0
1067 |  10
1068 | 13.986909
1069 |  20
1070 | 78.744863
1071 |  30
1072 | 0.0
1073 |  10
1074 | 14.482980
1075 |  20
1076 | 78.950313
1077 |  30
1078 | 0.0
1079 |  10
1080 | 15.000221
1081 |  20
1082 | 78.950260
1083 |  30
1084 | 0.0
1085 |   0
1086 | LINE
1087 |   5
1088 | 10b
1089 | 100
1090 | AcDbEntity
1091 |   8
1092 | Base
1093 |  62
1094 | 7
1095 | 100
1096 | AcDbLine
1097 |  10
1098 | 15.000221
1099 |  20
1100 | 78.950260
1101 |  30
1102 | 0.0
1103 |  11
1104 | 15.000221
1105 |  21
1106 | 78.950260
1107 |  31
1108 | 0.0
1109 |   0
1110 | SPLINE
1111 |   5
1112 | 10c
1113 | 100
1114 | AcDbEntity
1115 |   8
1116 | Base
1117 |  62
1118 | 7
1119 | 100
1120 | AcDbSpline
1121 |  70
1122 | 8
1123 |  71
1124 | 3
1125 |  72
1126 | 8
1127 |  73
1128 | 4
1129 |  74
1130 | 0
1131 |  40
1132 | 0
1133 |  40
1134 | 0
1135 |  40
1136 | 0
1137 |  40
1138 | 0
1139 |  40
1140 | 1
1141 |  40
1142 | 1
1143 |  40
1144 | 1
1145 |  40
1146 | 1
1147 |  10
1148 | 94.999748
1149 |  20
1150 | 78.950260
1151 |  30
1152 | 0.0
1153 |  10
1154 | 95.516989
1155 |  20
1156 | 78.950313
1157 |  30
1158 | 0.0
1159 |  10
1160 | 96.013061
1161 |  20
1162 | 78.744863
1163 |  30
1164 | 0.0
1165 |  10
1166 | 96.378806
1167 |  20
1168 | 78.379119
1169 |  30
1170 | 0.0
1171 |   0
1172 | SPLINE
1173 |   5
1174 | 10d
1175 | 100
1176 | AcDbEntity
1177 |   8
1178 | Base
1179 |  62
1180 | 7
1181 | 100
1182 | AcDbSpline
1183 |  70
1184 | 8
1185 |  71
1186 | 3
1187 |  72
1188 | 8
1189 |  73
1190 | 4
1191 |  74
1192 | 0
1193 |  40
1194 | 0
1195 |  40
1196 | 0
1197 |  40
1198 | 0
1199 |  40
1200 | 0
1201 |  40
1202 | 1
1203 |  40
1204 | 1
1205 |  40
1206 | 1
1207 |  40
1208 | 1
1209 |  10
1210 | 96.378806
1211 |  20
1212 | 78.379119
1213 |  30
1214 | 0.0
1215 |  10
1216 | 96.744551
1217 |  20
1218 | 78.013374
1219 |  30
1220 | 0.0
1221 |  10
1222 | 96.950001
1223 |  20
1224 | 77.517303
1225 |  30
1226 | 0.0
1227 |  10
1228 | 96.949949
1229 |  20
1230 | 77.000062
1231 |  30
1232 | 0.0
1233 |   0
1234 | SPLINE
1235 |   5
1236 | 10e
1237 | 100
1238 | AcDbEntity
1239 |   8
1240 | Base
1241 |  62
1242 | 7
1243 | 100
1244 | AcDbSpline
1245 |  70
1246 | 8
1247 |  71
1248 | 3
1249 |  72
1250 | 8
1251 |  73
1252 | 4
1253 |  74
1254 | 0
1255 |  40
1256 | 0
1257 |  40
1258 | 0
1259 |  40
1260 | 0
1261 |  40
1262 | 0
1263 |  40
1264 | 1
1265 |  40
1266 | 1
1267 |  40
1268 | 1
1269 |  40
1270 | 1
1271 |  10
1272 | 96.949949
1273 |  20
1274 | 77.000062
1275 |  30
1276 | 0.0
1277 |  10
1278 | 96.950002
1279 |  20
1280 | 76.482820
1281 |  30
1282 | 0.0
1283 |  10
1284 | 96.744552
1285 |  20
1286 | 75.986748
1287 |  30
1288 | 0.0
1289 |  10
1290 | 96.378807
1291 |  20
1292 | 75.621003
1293 |  30
1294 | 0.0
1295 |   0
1296 | SPLINE
1297 |   5
1298 | 10f
1299 | 100
1300 | AcDbEntity
1301 |   8
1302 | Base
1303 |  62
1304 | 7
1305 | 100
1306 | AcDbSpline
1307 |  70
1308 | 8
1309 |  71
1310 | 3
1311 |  72
1312 | 8
1313 |  73
1314 | 4
1315 |  74
1316 | 0
1317 |  40
1318 | 0
1319 |  40
1320 | 0
1321 |  40
1322 | 0
1323 |  40
1324 | 0
1325 |  40
1326 | 1
1327 |  40
1328 | 1
1329 |  40
1330 | 1
1331 |  40
1332 | 1
1333 |  10
1334 | 96.378807
1335 |  20
1336 | 75.621003
1337 |  30
1338 | 0.0
1339 |  10
1340 | 96.013062
1341 |  20
1342 | 75.255258
1343 |  30
1344 | 0.0
1345 |  10
1346 | 95.516990
1347 |  20
1348 | 75.049808
1349 |  30
1350 | 0.0
1351 |  10
1352 | 94.999748
1353 |  20
1354 | 75.049861
1355 |  30
1356 | 0.0
1357 |   0
1358 | SPLINE
1359 |   5
1360 | 110
1361 | 100
1362 | AcDbEntity
1363 |   8
1364 | Base
1365 |  62
1366 | 7
1367 | 100
1368 | AcDbSpline
1369 |  70
1370 | 8
1371 |  71
1372 | 3
1373 |  72
1374 | 8
1375 |  73
1376 | 4
1377 |  74
1378 | 0
1379 |  40
1380 | 0
1381 |  40
1382 | 0
1383 |  40
1384 | 0
1385 |  40
1386 | 0
1387 |  40
1388 | 1
1389 |  40
1390 | 1
1391 |  40
1392 | 1
1393 |  40
1394 | 1
1395 |  10
1396 | 94.999748
1397 |  20
1398 | 75.049861
1399 |  30
1400 | 0.0
1401 |  10
1402 | 93.922852
1403 |  20
1404 | 75.050056
1405 |  30
1406 | 0.0
1407 |  10
1408 | 93.049989
1409 |  20
1410 | 75.923166
1411 |  30
1412 | 0.0
1413 |  10
1414 | 93.050100
1415 |  20
1416 | 77.000062
1417 |  30
1418 | 0.0
1419 |   0
1420 | SPLINE
1421 |   5
1422 | 111
1423 | 100
1424 | AcDbEntity
1425 |   8
1426 | Base
1427 |  62
1428 | 7
1429 | 100
1430 | AcDbSpline
1431 |  70
1432 | 8
1433 |  71
1434 | 3
1435 |  72
1436 | 8
1437 |  73
1438 | 4
1439 |  74
1440 | 0
1441 |  40
1442 | 0
1443 |  40
1444 | 0
1445 |  40
1446 | 0
1447 |  40
1448 | 0
1449 |  40
1450 | 1
1451 |  40
1452 | 1
1453 |  40
1454 | 1
1455 |  40
1456 | 1
1457 |  10
1458 | 93.050100
1459 |  20
1460 | 77.000062
1461 |  30
1462 | 0.0
1463 |  10
1464 | 93.049991
1465 |  20
1466 | 78.076957
1467 |  30
1468 | 0.0
1469 |  10
1470 | 93.922853
1471 |  20
1472 | 78.950065
1473 |  30
1474 | 0.0
1475 |  10
1476 | 94.999748
1477 |  20
1478 | 78.950260
1479 |  30
1480 | 0.0
1481 |   0
1482 | LINE
1483 |   5
1484 | 112
1485 | 100
1486 | AcDbEntity
1487 |   8
1488 | Base
1489 |  62
1490 | 7
1491 | 100
1492 | AcDbLine
1493 |  10
1494 | 94.999748
1495 |  20
1496 | 78.950260
1497 |  30
1498 | 0.0
1499 |  11
1500 | 94.999748
1501 |  21
1502 | 78.950260
1503 |  31
1504 | 0.0
1505 |   0
1506 | SPLINE
1507 |   5
1508 | 113
1509 | 100
1510 | AcDbEntity
1511 |   8
1512 | Base
1513 |  62
1514 | 7
1515 | 100
1516 | AcDbSpline
1517 |  70
1518 | 8
1519 |  71
1520 | 3
1521 |  72
1522 | 8
1523 |  73
1524 | 4
1525 |  74
1526 | 0
1527 |  40
1528 | 0
1529 |  40
1530 | 0
1531 |  40
1532 | 0
1533 |  40
1534 | 0
1535 |  40
1536 | 1
1537 |  40
1538 | 1
1539 |  40
1540 | 1
1541 |  40
1542 | 1
1543 |  10
1544 | 15.000221
1545 |  20
1546 | 14.949758
1547 |  30
1548 | 0.0
1549 |  10
1550 | 16.076900
1551 |  20
1552 | 14.949564
1553 |  30
1554 | 0.0
1555 |  10
1556 | 16.949672
1557 |  20
1558 | 14.076795
1559 |  30
1560 | 0.0
1561 |  10
1562 | 16.949870
1563 |  20
1564 | 13.000116
1565 |  30
1566 | 0.0
1567 |   0
1568 | SPLINE
1569 |   5
1570 | 114
1571 | 100
1572 | AcDbEntity
1573 |   8
1574 | Base
1575 |  62
1576 | 7
1577 | 100
1578 | AcDbSpline
1579 |  70
1580 | 8
1581 |  71
1582 | 3
1583 |  72
1584 | 8
1585 |  73
1586 | 4
1587 |  74
1588 | 0
1589 |  40
1590 | 0
1591 |  40
1592 | 0
1593 |  40
1594 | 0
1595 |  40
1596 | 0
1597 |  40
1598 | 1
1599 |  40
1600 | 1
1601 |  40
1602 | 1
1603 |  40
1604 | 1
1605 |  10
1606 | 16.949870
1607 |  20
1608 | 13.000116
1609 |  30
1610 | 0.0
1611 |  10
1612 | 16.949987
1613 |  20
1614 | 11.923216
1615 |  30
1616 | 0.0
1617 |  10
1618 | 16.077122
1619 |  20
1620 | 11.050099
1621 |  30
1622 | 0.0
1623 |  10
1624 | 15.000221
1625 |  20
1626 | 11.049904
1627 |  30
1628 | 0.0
1629 |   0
1630 | SPLINE
1631 |   5
1632 | 115
1633 | 100
1634 | AcDbEntity
1635 |   8
1636 | Base
1637 |  62
1638 | 7
1639 | 100
1640 | AcDbSpline
1641 |  70
1642 | 8
1643 |  71
1644 | 3
1645 |  72
1646 | 8
1647 |  73
1648 | 4
1649 |  74
1650 | 0
1651 |  40
1652 | 0
1653 |  40
1654 | 0
1655 |  40
1656 | 0
1657 |  40
1658 | 0
1659 |  40
1660 | 1
1661 |  40
1662 | 1
1663 |  40
1664 | 1
1665 |  40
1666 | 1
1667 |  10
1668 | 15.000221
1669 |  20
1670 | 11.049904
1671 |  30
1672 | 0.0
1673 |  10
1674 | 14.482978
1675 |  20
1676 | 11.049851
1677 |  30
1678 | 0.0
1679 |  10
1680 | 13.986904
1681 |  20
1682 | 11.255303
1683 |  30
1684 | 0.0
1685 |  10
1686 | 13.621159
1687 |  20
1688 | 11.621050
1689 |  30
1690 | 0.0
1691 |   0
1692 | SPLINE
1693 |   5
1694 | 116
1695 | 100
1696 | AcDbEntity
1697 |   8
1698 | Base
1699 |  62
1700 | 7
1701 | 100
1702 | AcDbSpline
1703 |  70
1704 | 8
1705 |  71
1706 | 3
1707 |  72
1708 | 8
1709 |  73
1710 | 4
1711 |  74
1712 | 0
1713 |  40
1714 | 0
1715 |  40
1716 | 0
1717 |  40
1718 | 0
1719 |  40
1720 | 0
1721 |  40
1722 | 1
1723 |  40
1724 | 1
1725 |  40
1726 | 1
1727 |  40
1728 | 1
1729 |  10
1730 | 13.621159
1731 |  20
1732 | 11.621050
1733 |  30
1734 | 0.0
1735 |  10
1736 | 13.255414
1737 |  20
1738 | 11.986798
1739 |  30
1740 | 0.0
1741 |  10
1742 | 13.049965
1743 |  20
1744 | 12.482873
1745 |  30
1746 | 0.0
1747 |  10
1748 | 13.050022
1749 |  20
1750 | 13.000116
1751 |  30
1752 | 0.0
1753 |   0
1754 | SPLINE
1755 |   5
1756 | 117
1757 | 100
1758 | AcDbEntity
1759 |   8
1760 | Base
1761 |  62
1762 | 7
1763 | 100
1764 | AcDbSpline
1765 |  70
1766 | 8
1767 |  71
1768 | 3
1769 |  72
1770 | 8
1771 |  73
1772 | 4
1773 |  74
1774 | 0
1775 |  40
1776 | 0
1777 |  40
1778 | 0
1779 |  40
1780 | 0
1781 |  40
1782 | 0
1783 |  40
1784 | 1
1785 |  40
1786 | 1
1787 |  40
1788 | 1
1789 |  40
1790 | 1
1791 |  10
1792 | 13.050022
1793 |  20
1794 | 13.000116
1795 |  30
1796 | 0.0
1797 |  10
1798 | 13.050219
1799 |  20
1800 | 14.077010
1801 |  30
1802 | 0.0
1803 |  10
1804 | 13.923328
1805 |  20
1806 | 14.949869
1807 |  30
1808 | 0.0
1809 |  10
1810 | 15.000221
1811 |  20
1812 | 14.949758
1813 |  30
1814 | 0.0
1815 |   0
1816 | LINE
1817 |   5
1818 | 118
1819 | 100
1820 | AcDbEntity
1821 |   8
1822 | Base
1823 |  62
1824 | 7
1825 | 100
1826 | AcDbLine
1827 |  10
1828 | 15.000221
1829 |  20
1830 | 14.949758
1831 |  30
1832 | 0.0
1833 |  11
1834 | 15.000221
1835 |  21
1836 | 14.949758
1837 |  31
1838 | 0.0
1839 |   0
1840 | SPLINE
1841 |   5
1842 | 119
1843 | 100
1844 | AcDbEntity
1845 |   8
1846 | Base
1847 |  62
1848 | 7
1849 | 100
1850 | AcDbSpline
1851 |  70
1852 | 8
1853 |  71
1854 | 3
1855 |  72
1856 | 8
1857 |  73
1858 | 4
1859 |  74
1860 | 0
1861 |  40
1862 | 0
1863 |  40
1864 | 0
1865 |  40
1866 | 0
1867 |  40
1868 | 0
1869 |  40
1870 | 1
1871 |  40
1872 | 1
1873 |  40
1874 | 1
1875 |  40
1876 | 1
1877 |  10
1878 | 94.999748
1879 |  20
1880 | 14.949758
1881 |  30
1882 | 0.0
1883 |  10
1884 | 96.076642
1885 |  20
1886 | 14.949869
1887 |  30
1888 | 0.0
1889 |  10
1890 | 96.949751
1891 |  20
1892 | 14.077010
1893 |  30
1894 | 0.0
1895 |  10
1896 | 96.949949
1897 |  20
1898 | 13.000116
1899 |  30
1900 | 0.0
1901 |   0
1902 | SPLINE
1903 |   5
1904 | 11a
1905 | 100
1906 | AcDbEntity
1907 |   8
1908 | Base
1909 |  62
1910 | 7
1911 | 100
1912 | AcDbSpline
1913 |  70
1914 | 8
1915 |  71
1916 | 3
1917 |  72
1918 | 8
1919 |  73
1920 | 4
1921 |  74
1922 | 0
1923 |  40
1924 | 0
1925 |  40
1926 | 0
1927 |  40
1928 | 0
1929 |  40
1930 | 0
1931 |  40
1932 | 1
1933 |  40
1934 | 1
1935 |  40
1936 | 1
1937 |  40
1938 | 1
1939 |  10
1940 | 96.949949
1941 |  20
1942 | 13.000116
1943 |  30
1944 | 0.0
1945 |  10
1946 | 96.950005
1947 |  20
1948 | 12.482873
1949 |  30
1950 | 0.0
1951 |  10
1952 | 96.744556
1953 |  20
1954 | 11.986798
1955 |  30
1956 | 0.0
1957 |  10
1958 | 96.378811
1959 |  20
1960 | 11.621050
1961 |  30
1962 | 0.0
1963 |   0
1964 | SPLINE
1965 |   5
1966 | 11b
1967 | 100
1968 | AcDbEntity
1969 |   8
1970 | Base
1971 |  62
1972 | 7
1973 | 100
1974 | AcDbSpline
1975 |  70
1976 | 8
1977 |  71
1978 | 3
1979 |  72
1980 | 8
1981 |  73
1982 | 4
1983 |  74
1984 | 0
1985 |  40
1986 | 0
1987 |  40
1988 | 0
1989 |  40
1990 | 0
1991 |  40
1992 | 0
1993 |  40
1994 | 1
1995 |  40
1996 | 1
1997 |  40
1998 | 1
1999 |  40
2000 | 1
2001 |  10
2002 | 96.378811
2003 |  20
2004 | 11.621050
2005 |  30
2006 | 0.0
2007 |  10
2008 | 96.013065
2009 |  20
2010 | 11.255303
2011 |  30
2012 | 0.0
2013 |  10
2014 | 95.516992
2015 |  20
2016 | 11.049851
2017 |  30
2018 | 0.0
2019 |  10
2020 | 94.999748
2021 |  20
2022 | 11.049904
2023 |  30
2024 | 0.0
2025 |   0
2026 | SPLINE
2027 |   5
2028 | 11c
2029 | 100
2030 | AcDbEntity
2031 |   8
2032 | Base
2033 |  62
2034 | 7
2035 | 100
2036 | AcDbSpline
2037 |  70
2038 | 8
2039 |  71
2040 | 3
2041 |  72
2042 | 8
2043 |  73
2044 | 4
2045 |  74
2046 | 0
2047 |  40
2048 | 0
2049 |  40
2050 | 0
2051 |  40
2052 | 0
2053 |  40
2054 | 0
2055 |  40
2056 | 1
2057 |  40
2058 | 1
2059 |  40
2060 | 1
2061 |  40
2062 | 1
2063 |  10
2064 | 94.999748
2065 |  20
2066 | 11.049904
2067 |  30
2068 | 0.0
2069 |  10
2070 | 93.922848
2071 |  20
2072 | 11.050099
2073 |  30
2074 | 0.0
2075 |  10
2076 | 93.049983
2077 |  20
2078 | 11.923216
2079 |  30
2080 | 0.0
2081 |  10
2082 | 93.050100
2083 |  20
2084 | 13.000116
2085 |  30
2086 | 0.0
2087 |   0
2088 | SPLINE
2089 |   5
2090 | 11d
2091 | 100
2092 | AcDbEntity
2093 |   8
2094 | Base
2095 |  62
2096 | 7
2097 | 100
2098 | AcDbSpline
2099 |  70
2100 | 8
2101 |  71
2102 | 3
2103 |  72
2104 | 8
2105 |  73
2106 | 4
2107 |  74
2108 | 0
2109 |  40
2110 | 0
2111 |  40
2112 | 0
2113 |  40
2114 | 0
2115 |  40
2116 | 0
2117 |  40
2118 | 1
2119 |  40
2120 | 1
2121 |  40
2122 | 1
2123 |  40
2124 | 1
2125 |  10
2126 | 93.050100
2127 |  20
2128 | 13.000116
2129 |  30
2130 | 0.0
2131 |  10
2132 | 93.050298
2133 |  20
2134 | 14.076794
2135 |  30
2136 | 0.0
2137 |  10
2138 | 93.923070
2139 |  20
2140 | 14.949564
2141 |  30
2142 | 0.0
2143 |  10
2144 | 94.999748
2145 |  20
2146 | 14.949758
2147 |  30
2148 | 0.0
2149 |   0
2150 | LINE
2151 |   5
2152 | 11e
2153 | 100
2154 | AcDbEntity
2155 |   8
2156 | Base
2157 |  62
2158 | 7
2159 | 100
2160 | AcDbLine
2161 |  10
2162 | 94.999748
2163 |  20
2164 | 14.949758
2165 |  30
2166 | 0.0
2167 |  11
2168 | 94.999748
2169 |  21
2170 | 14.949758
2171 |  31
2172 | 0.0
2173 |   0
2174 | ENDSEC
2175 |   0
2176 | SECTION
2177 |   2
2178 | OBJECTS
2179 |   0
2180 | DICTIONARY
2181 |   5
2182 | C
2183 | 330
2184 | 0
2185 | 100
2186 | AcDbDictionary
2187 |   3
2188 | ACAD_GROUP
2189 | 350
2190 | D
2191 |   3
2192 | ACAD_MLINESTYLE
2193 | 350
2194 | 17
2195 |   0
2196 | DICTIONARY
2197 |   5
2198 | D
2199 | 330
2200 | C
2201 | 100
2202 | AcDbDictionary
2203 |   0
2204 | DICTIONARY
2205 |   5
2206 | 1A
2207 | 330
2208 | C
2209 | 100
2210 | AcDbDictionary
2211 |   0
2212 | DICTIONARY
2213 |   5
2214 | 17
2215 | 330
2216 | C
2217 | 100
2218 | AcDbDictionary
2219 |   3
2220 | STANDARD
2221 | 350
2222 | 18
2223 |   0
2224 | DICTIONARY
2225 |   5
2226 | 19
2227 | 330
2228 | C
2229 | 100
2230 | AcDbDictionary
2231 |   0
2232 | ENDSEC
2233 |   0
2234 | EOF
2235 | 


--------------------------------------------------------------------------------
/hardware/plate.gcode:
--------------------------------------------------------------------------------
 1 | M5
 2 | G21 G90 G64 G40
 3 | G0 Z3.0
 4 | T0 M6
 5 | G17
 6 | G0 X18.571 Y98.3292
 7 | G0 Z1.0
 8 | G1 F300.0 Z-0.4
 9 | M3 S100
10 | G2 F100.0 X19.9501 Y98.9004 I1.3789 J-1.3789
11 | G2 X21.8997 Y96.9502 I-0.0004 J-1.95
12 | G2 X19.9501 Y95.0 I-1.95 J-0.0002
13 | G2 X18.571 Y95.5711 I-0.0002 J1.95
14 | G2 X17.9999 Y96.9502 I1.3789 J1.3789
15 | G2 X18.571 Y98.3292 I1.95 J0.0002
16 | M5
17 | G0 Z3.0
18 | G0 X19.9087 Y34.8995
19 | G0 Z1.0
20 | G1 F300.0 Z-0.4
21 | M3 S100
22 | G2 F100.0 X19.9501 Y34.8999 I0.0411 J-1.9496
23 | G2 X21.8997 Y32.9502 I-0.0004 J-1.95
24 | G2 X19.9501 Y31.0 I-1.95 J-0.0002
25 | G2 X18.571 Y31.5712 I-0.0002 J1.95
26 | G2 X17.9999 Y32.9502 I1.3788 J1.3788
27 | G2 X19.9087 Y34.8995 I1.95 J-0.0004
28 | M5
29 | G0 Z3.0
30 | G0 X98.0005 Y32.9974
31 | G0 Z1.0
32 | G1 F300.0 Z-0.4
33 | M3 S100
34 | G2 F100.0 X99.9496 Y34.8999 I1.9494 J-0.0475
35 | G2 X101.8998 Y32.9502 I0.0002 J-1.95
36 | G2 X101.3287 Y31.5712 I-1.95 J-0.0002
37 | G2 X99.9496 Y31.0 I-1.3789 J1.3789
38 | G2 X97.9999 Y32.9502 I0.0004 J1.95
39 | G2 X98.0005 Y32.9974 I1.95 J-0.0004
40 | M5
41 | G0 Z3.0
42 | G0 X99.8905 Y95.0009
43 | G0 Z1.0
44 | G1 F300.0 Z-0.4
45 | M3 S100
46 | G2 F100.0 X97.9999 Y96.9502 I0.0594 J1.9491
47 | G2 X99.9496 Y98.9004 I1.95 J0.0002
48 | G2 X101.3286 Y98.3292 I0.0002 J-1.95
49 | G2 X101.8998 Y96.9502 I-1.3788 J-1.3789
50 | G2 X101.3286 Y95.5711 I-1.95 J-0.0002
51 | G2 X99.9496 Y95.0 I-1.3789 J1.3789
52 | G2 X99.8905 Y95.0009 I0.0004 J1.95
53 | M5
54 | G0 Z3.0
55 | G0 X5.0 Y109.8999
56 | G0 Z1.0
57 | G1 F300.0 Z-0.2
58 | M3 S90
59 | G1 F200.0 Y20.0
60 | G1 X119.8997
61 | G1 Y109.8999
62 | G1 X5.0
63 | G1 F300.0 Z-0.4
64 | G1 F200.0 Y20.0
65 | G1 X119.8997
66 | G1 Y109.8999
67 | G1 X5.0
68 | M5
69 | G0 Z3.0
70 | G28
71 | M30
72 | 


--------------------------------------------------------------------------------
/hardware/plate.svg:
--------------------------------------------------------------------------------
 1 | <?xml version="1.0" encoding="UTF-8" standalone="no"?>
 2 | <!-- Created with Inkscape (http://www.inkscape.org/) -->
 3 | 
 4 | <svg
 5 |    xmlns:dc="http://purl.org/dc/elements/1.1/"
 6 |    xmlns:cc="http://creativecommons.org/ns#"
 7 |    xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
 8 |    xmlns:svg="http://www.w3.org/2000/svg"
 9 |    xmlns="http://www.w3.org/2000/svg"
10 |    xmlns:sodipodi="http://sodipodi.sourceforge.net/DTD/sodipodi-0.dtd"
11 |    xmlns:inkscape="http://www.inkscape.org/namespaces/inkscape"
12 |    width="210mm"
13 |    height="297mm"
14 |    viewBox="0 0 744.09448819 1052.3622047"
15 |    id="svg2"
16 |    version="1.1"
17 |    inkscape:version="0.91 r13725"
18 |    sodipodi:docname="plate.svg">
19 |   <defs
20 |      id="defs4" />
21 |   <sodipodi:namedview
22 |      id="base"
23 |      pagecolor="#c8c8c8"
24 |      bordercolor="#666666"
25 |      borderopacity="1.0"
26 |      inkscape:pageopacity="0"
27 |      inkscape:pageshadow="2"
28 |      inkscape:zoom="1.4452285"
29 |      inkscape:cx="87.115746"
30 |      inkscape:cy="159.44883"
31 |      inkscape:document-units="mm"
32 |      inkscape:current-layer="g4212"
33 |      showgrid="false"
34 |      inkscape:window-width="1440"
35 |      inkscape:window-height="851"
36 |      inkscape:window-x="0"
37 |      inkscape:window-y="1"
38 |      inkscape:window-maximized="1" />
39 |   <metadata
40 |      id="metadata7">
41 |     <rdf:RDF>
42 |       <cc:Work
43 |          rdf:about="">
44 |         <dc:format>image/svg+xml</dc:format>
45 |         <dc:type
46 |            rdf:resource="http://purl.org/dc/dcmitype/StillImage" />
47 |         <dc:title></dc:title>
48 |       </cc:Work>
49 |     </rdf:RDF>
50 |   </metadata>
51 |   <g
52 |      inkscape:label="Base"
53 |      inkscape:groupmode="layer"
54 |      id="layer1">
55 |     <ellipse
56 |        style="fill:none;fill-opacity:0.50196078;stroke:none;stroke-width:0.35433072;stroke-miterlimit:4;stroke-dasharray:none"
57 |        id="path4208"
58 |        cx="-201.36765"
59 |        cy="884.41302"
60 |        rx="21.422091"
61 |        ry="12.853253" />
62 |     <g
63 |        id="g4212">
64 |       <path
65 |          style="fill:#0000ff;fill-opacity:0.50196081;stroke:#000000;stroke-width:0.35433071;stroke-miterlimit:4;stroke-dasharray:none"
66 |          d="M 0.17773438 733.64258 L 0.17773438 1052.1855 L 407.30273 1052.1855 L 407.30273 733.64258 L 0.17773438 733.64258 z M 53.150391 772.61719 A 6.9094486 6.9094491 0 0 1 60.058594 779.52734 A 6.9094486 6.9094491 0 0 1 53.150391 786.4375 A 6.9094486 6.9094491 0 0 1 46.240234 779.52734 A 6.9094486 6.9094491 0 0 1 53.150391 772.61719 z M 336.61328 772.61719 A 6.9094486 6.9094486 0 0 1 343.52344 779.52734 A 6.9094486 6.9094486 0 0 1 336.61328 786.4375 A 6.9094486 6.9094486 0 0 1 329.70508 779.52734 A 6.9094486 6.9094486 0 0 1 336.61328 772.61719 z M 53.150391 999.39062 A 6.9094486 6.9094486 0 0 1 60.058594 1006.2988 A 6.9094486 6.9094486 0 0 1 53.150391 1013.209 A 6.9094486 6.9094486 0 0 1 46.240234 1006.2988 A 6.9094486 6.9094486 0 0 1 53.150391 999.39062 z M 336.61328 999.39062 A 6.9094486 6.9094491 0 0 1 343.52344 1006.2988 A 6.9094486 6.9094491 0 0 1 336.61328 1013.209 A 6.9094486 6.9094491 0 0 1 329.70508 1006.2988 A 6.9094486 6.9094491 0 0 1 336.61328 999.39062 z "
67 |          id="rect4136" />
68 |     </g>
69 |   </g>
70 | </svg>
71 | 


--------------------------------------------------------------------------------
/readme.markdown:
--------------------------------------------------------------------------------
 1 | NodeBot AI project(s)
 2 | =====================
 3 | 
 4 | Contents
 5 | --------
 6 | - [Hardware](#hardware)
 7 | - [Hardware assembly](#hardware-assembly)
 8 | - [Hardware power supply](#hardware-power-supply)
 9 | - [Software installation](#software-installation)
10 | 
11 | <a name="hardware" />
12 | Hardware
13 | --------
14 | 
15 | <a name="hardware-assembly" />
16 | Hardware assembly
17 | -----------------
18 | 
19 | See [Hardware directory](hardware)
20 | 
21 | <a name="hardware-power-supply" />
22 | Hardware power supply
23 | ---------------------
24 | 
25 | All measurements at 5 VDC.  Raspberry Pi 2 has Wi-Pi (USB Wi-Fi dongle).
26 | 
27 | - mBot idle (with LED matrix, line following, etc): 0.15 mA
28 | - mBot motors running (@ 100 out of 255): 0.21 mA
29 | - mBot motors starting/spinning up: 0.27 mA
30 | - Pixy camera starting up: 0.25 mA
31 | - Pixy camera idle: 0.18 mA
32 | - RPi2 starting up: 0.38 mA
33 | - RPi2 idle: 0.28 mA
34 | - RPi2 Wi-Fi data transfer: 0.38 mA to 0.44 mA
35 | - RPi2 capturing video: 0.46 mA (note: CPU is idling)
36 | - RPi2 capturing video and Wi-Fi data transfer: 0.66 mA
37 | - RPi2 one CPU thread busy loop (25%): 0.31 mA
38 | - RPi2 four CPU threads busy loop (100%): 0.41 mA
39 | - RPi2 CPU 100%, video and Wi-Fi transfer: 0.76 mA
40 | 
41 | All up: RPi2 0.76 mA + Pixy 0.25 mA + mBot 0.27 mA = 1.28 mA total
42 | 
43 | <a name="software-installation"/>
44 | ## Software installation
45 | 
46 | ### Node:
47 | 
48 | * Make sure NodeJS is installed (pref V6 branch)
49 | * cd to src/nodebot and run `npm install`
50 | 
51 | ### mBot Firmware
52 | 
53 | Ensure `./node_modules/.bin` is on $PATH then:
54 | 
55 | ```
56 | interchange install git+https://github.com/Makeblock-official/mbot_nodebots -a uno --firmata=usb
57 | ```
58 | 
59 | 
60 | 
61 | 


--------------------------------------------------------------------------------
/src/ai/camera_capture.sh:
--------------------------------------------------------------------------------
 1 | #!/bin/sh
 2 | #
 3 | # Local video capture sent over GStreamer RTP/H.264/UDP to local/remote hosts
 4 | 
 5 | REMOTE_HOST=nomad.local  # classify_video.py
 6 | REMOTE_PORT=5001         #   "           "
 7 | 
 8 | if [ $# -ge 1 ]; then
 9 |   REMOTE_HOST=$1
10 | fi
11 | 
12 | if [ $# -ge 2 ]; then
13 |   REMOTE_PORT=$2
14 | fi
15 | 
16 | # Raspberry Pi camera
17 | #
18 | BIT_RATE=1000000
19 | FPS=10
20 | # Ratio 16:9 1.777
21 | WIDTH=512
22 | HEIGHT=288
23 | 
24 | # Local host UDP stream
25 | #
26 | HOST_L=127.0.0.1
27 | PORT_L=5000
28 | FPS_L=10/1
29 | # Ratio: 1:1
30 | WIDTH_L=128
31 | HEIGHT_L=128
32 | 
33 | # Remote host UDP stream
34 | #
35 | HOST_R=$REMOTE_HOST
36 | PORT_R=$REMOTE_PORT
37 | FPS_R=10/1
38 | # Ratio 16:9 1.777
39 | WIDTH_R=512
40 | HEIGHT_R=288
41 | 
42 | # If remote host can't be found, then stream to localhost
43 | #
44 | ping -c 1 -q $HOST_R >/dev/null 2>&1
45 | 
46 | if [ $? != 0 ]; then
47 |   HOST_R=127.0.0.1
48 | fi
49 | 
50 | echo Remote camera stream: $HOST_R:$PORT_R
51 | 
52 | OS=`uname`
53 | 
54 | if [ $OS = "Darwin" ]; then
55 |   H264_DECODE=avdec_h264
56 | # H264_ENCODE=x264enc 
57 | # H264_ENCODE=avenc_mpeg4   # software encoder
58 |   H264_ENCODE=vtenc_h264
59 | 
60 |   OVERLAY='queue'
61 | else
62 |   H264_DECODE=omxh264dec
63 |   H264_ENCODE=omxh264enc
64 | 
65 |   OVERLAY='timeoverlay halignment=right valignment=top text="Elapsed: " shaded-background=true ! clockoverlay halignment=left valignment=top time-format="%Y/%m/%d %H:%M:%S" shaded-background=true'
66 | fi
67 | 
68 | grep -q BCM270. /proc/cpuinfo >/dev/null 2>&1
69 | 
70 | if [ $? = 0 ]; then
71 |   RASPIVID="raspivid --flush -hf -vf -n -t 0 -b $BIT_RATE -fps $FPS -w $WIDTH -h $HEIGHT -o -"
72 |   CAMERA_SOURCE="fdsrc ! h264parse ! $H264_DECODE"
73 | else
74 |   RASPIVID="echo"
75 |   CAMERA_SOURCE=autovideosrc
76 | fi
77 | 
78 | $RASPIVID | gst-launch-1.0 $CAMERA_SOURCE ! \
79 |   tee name=tee_local ! queue ! \
80 |     videoscale !  videorate  ! videoconvert ! \
81 |     video/x-raw,width=$WIDTH_L,height=$HEIGHT_L,framerate=$FPS_L ! \
82 |     $H264_ENCODE ! \
83 |     rtph264pay config-interval=5 pt=96 ! \
84 |     udpsink host=$HOST_L port=$PORT_L sync=false async=true \
85 |     tee_local. ! queue ! \
86 |   tee name=tee_remote ! queue ! \
87 |     videoscale !  videorate  ! videoconvert ! \
88 |     video/x-raw,width=$WIDTH_R,height=$HEIGHT_R,framerate=$FPS_R ! \
89 |     $OVERLAY ! \
90 |     $H264_ENCODE ! \
91 |     video/x-h264,width=$WIDTH_R,height=$HEIGHT_R,framerate=$FPS_R ! \
92 |     rtph264pay config-interval=5 pt=96 ! \
93 |     udpsink host=$HOST_R port=$PORT_R sync=false async=true
94 | 


--------------------------------------------------------------------------------
/src/ai/camera_watch.sh:
--------------------------------------------------------------------------------
 1 | #!/bin/sh
 2 | #
 3 | # Watch remote GStreamer RTP/H.264/UDP video stream
 4 | 
 5 | LOCAL_PORT=5001
 6 | 
 7 | if [ $# = 1 ]; then
 8 |   LOCAL_PORT=$1
 9 | fi
10 | 
11 | # sync=f: Play as soon as data arrives
12 | # sync=t: Play depending upon video buffer timestamps
13 | #
14 | VIDEO_SYNC="sync=f"
15 | VIDEO_SINK=ximagesink
16 | 
17 | OS=`uname`
18 | 
19 | if [ $OS = "Darwin" ]; then
20 |   VIDEO_SINK=osxvideosink
21 | fi
22 | 
23 | gst-launch-1.0 udpsrc port=$LOCAL_PORT \
24 |   caps='application/x-rtp, media=(string)video, clock-rate=(int)90000, encoding-name=(string)H264' ! \
25 |   rtph264depay ! avdec_h264 ! videoconvert ! $VIDEO_SINK $VIDEO_SYNC
26 | 


--------------------------------------------------------------------------------
/src/ai/explore_generate_data.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 1,
  6 |    "metadata": {
  7 |     "collapsed": false
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "import numpy as np\n",
 12 |     "from skimage.draw import polygon"
 13 |    ]
 14 |   },
 15 |   {
 16 |    "cell_type": "code",
 17 |    "execution_count": 2,
 18 |    "metadata": {
 19 |     "collapsed": false
 20 |    },
 21 |    "outputs": [],
 22 |    "source": [
 23 |     "import matplotlib.pyplot as plt\n",
 24 |     "%matplotlib inline"
 25 |    ]
 26 |   },
 27 |   {
 28 |    "cell_type": "code",
 29 |    "execution_count": 3,
 30 |    "metadata": {
 31 |     "collapsed": false
 32 |    },
 33 |    "outputs": [],
 34 |    "source": [
 35 |     "%load_ext autoreload\n",
 36 |     "%autoreload 1\n",
 37 |     "%aimport generate_data\n",
 38 |     "from generate_data import *"
 39 |    ]
 40 |   },
 41 |   {
 42 |    "cell_type": "code",
 43 |    "execution_count": 4,
 44 |    "metadata": {
 45 |     "collapsed": false
 46 |    },
 47 |    "outputs": [
 48 |     {
 49 |      "data": {
 50 |       "text/plain": [
 51 |        "<matplotlib.image.AxesImage at 0x112b20190>"
 52 |       ]
 53 |      },
 54 |      "execution_count": 4,
 55 |      "metadata": {},
 56 |      "output_type": "execute_result"
 57 |     },
 58 |     {
 59 |      "data": {
 60 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP4AAAD+CAYAAAAalrhRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
 61 |       "text/plain": [
 62 |        "<matplotlib.figure.Figure at 0x10f768890>"
 63 |       ]
 64 |      },
 65 |      "metadata": {},
 66 |      "output_type": "display_data"
 67 |     }
 68 |    ],
 69 |    "source": [
 70 |     "plt.imshow(generate_data.create_road(64, 64, offset = -21)[:,:,0], interpolation='nearest')"
 71 |    ]
 72 |   },
 73 |   {
 74 |    "cell_type": "code",
 75 |    "execution_count": 5,
 76 |    "metadata": {
 77 |     "collapsed": false
 78 |    },
 79 |    "outputs": [
 80 |     {
 81 |      "name": "stdout",
 82 |      "output_type": "stream",
 83 |      "text": [
 84 |       "8.44379322716\n"
 85 |      ]
 86 |     },
 87 |     {
 88 |      "data": {
 89 |       "text/plain": [
 90 |        "<matplotlib.image.AxesImage at 0x112d2ead0>"
 91 |       ]
 92 |      },
 93 |      "execution_count": 5,
 94 |      "metadata": {},
 95 |      "output_type": "execute_result"
 96 |     },
 97 |     {
 98 |      "data": {
 99 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP4AAAD+CAYAAAAalrhRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
100 |       "text/plain": [
101 |        "<matplotlib.figure.Figure at 0x112b6fa90>"
102 |       ]
103 |      },
104 |      "metadata": {},
105 |      "output_type": "display_data"
106 |     }
107 |    ],
108 |    "source": [
109 |     "image, offset = generate_data.random_road(64, 64)\n",
110 |     "print(offset)\n",
111 |     "plt.imshow(image[:,:,0])"
112 |    ]
113 |   },
114 |   {
115 |    "cell_type": "code",
116 |    "execution_count": 6,
117 |    "metadata": {
118 |     "collapsed": false
119 |    },
120 |    "outputs": [
121 |     {
122 |      "name": "stdout",
123 |      "output_type": "stream",
124 |      "text": [
125 |       "[ 0.06889187  0.20798574]\n"
126 |      ]
127 |     },
128 |     {
129 |      "data": {
130 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP4AAAD+CAYAAAAalrhRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
131 |       "text/plain": [
132 |        "<matplotlib.figure.Figure at 0x10f7b7750>"
133 |       ]
134 |      },
135 |      "metadata": {},
136 |      "output_type": "display_data"
137 |     },
138 |     {
139 |      "data": {
140 |       "text/plain": [
141 |        "<matplotlib.image.AxesImage at 0x112f8e050>"
142 |       ]
143 |      },
144 |      "execution_count": 6,
145 |      "metadata": {},
146 |      "output_type": "execute_result"
147 |     },
148 |     {
149 |      "data": {
150 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP4AAAD+CAYAAAAalrhRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n",
151 |       "text/plain": [
152 |        "<matplotlib.figure.Figure at 0x112bded10>"
153 |       ]
154 |      },
155 |      "metadata": {},
156 |      "output_type": "display_data"
157 |     }
158 |    ],
159 |    "source": [
160 |     "input_data, label_data = generate_data.generate_batch(64, 64, 2)\n",
161 |     "print(label_data)\n",
162 |     "plt.imshow(input_data[0,:,:,0])\n",
163 |     "plt.show()\n",
164 |     "plt.imshow(input_data[1,:,:,0])"
165 |    ]
166 |   },
167 |   {
168 |    "cell_type": "code",
169 |    "execution_count": null,
170 |    "metadata": {
171 |     "collapsed": true
172 |    },
173 |    "outputs": [],
174 |    "source": []
175 |   }
176 |  ],
177 |  "metadata": {
178 |   "kernelspec": {
179 |    "display_name": "Python 2",
180 |    "language": "python",
181 |    "name": "python2"
182 |   },
183 |   "language_info": {
184 |    "codemirror_mode": {
185 |     "name": "ipython",
186 |     "version": 2
187 |    },
188 |    "file_extension": ".py",
189 |    "mimetype": "text/x-python",
190 |    "name": "python",
191 |    "nbconvert_exporter": "python",
192 |    "pygments_lexer": "ipython2",
193 |    "version": "2.7.12"
194 |   }
195 |  },
196 |  "nbformat": 4,
197 |  "nbformat_minor": 0
198 | }
199 | 


--------------------------------------------------------------------------------
/src/ai/explore_run.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 1,
  6 |    "metadata": {
  7 |     "collapsed": true
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "import numpy as np"
 12 |    ]
 13 |   },
 14 |   {
 15 |    "cell_type": "code",
 16 |    "execution_count": 2,
 17 |    "metadata": {
 18 |     "collapsed": true
 19 |    },
 20 |    "outputs": [],
 21 |    "source": [
 22 |     "import tensorflow as tf\n",
 23 |     "import tflearn"
 24 |    ]
 25 |   },
 26 |   {
 27 |    "cell_type": "code",
 28 |    "execution_count": 3,
 29 |    "metadata": {
 30 |     "collapsed": false
 31 |    },
 32 |    "outputs": [],
 33 |    "source": [
 34 |     "%reload_ext autoreload\n",
 35 |     "%autoreload 1\n",
 36 |     "%aimport train\n",
 37 |     "%aimport generate_data"
 38 |    ]
 39 |   },
 40 |   {
 41 |    "cell_type": "code",
 42 |    "execution_count": 4,
 43 |    "metadata": {
 44 |     "collapsed": false
 45 |    },
 46 |    "outputs": [],
 47 |    "source": [
 48 |     "tf.reset_default_graph()\n",
 49 |     "model = train.build_model()"
 50 |    ]
 51 |   },
 52 |   {
 53 |    "cell_type": "code",
 54 |    "execution_count": 68,
 55 |    "metadata": {
 56 |     "collapsed": false
 57 |    },
 58 |    "outputs": [],
 59 |    "source": [
 60 |     "model.load('checkpoints/road_model1-800')"
 61 |    ]
 62 |   },
 63 |   {
 64 |    "cell_type": "code",
 65 |    "execution_count": 69,
 66 |    "metadata": {
 67 |     "collapsed": false
 68 |    },
 69 |    "outputs": [],
 70 |    "source": [
 71 |     "input_size = 128\n",
 72 |     "data_size = 1\n",
 73 |     "\n",
 74 |     "image_data, label_data = \\\n",
 75 |     "    generate_data.generate_batch(\n",
 76 |     "        height=input_size,\n",
 77 |     "        width=input_size,\n",
 78 |     "        minibatch_size=data_size)"
 79 |    ]
 80 |   },
 81 |   {
 82 |    "cell_type": "code",
 83 |    "execution_count": 70,
 84 |    "metadata": {
 85 |     "collapsed": false
 86 |    },
 87 |    "outputs": [],
 88 |    "source": [
 89 |     "pr = model.predict(image_data)"
 90 |    ]
 91 |   },
 92 |   {
 93 |    "cell_type": "code",
 94 |    "execution_count": 71,
 95 |    "metadata": {
 96 |     "collapsed": true
 97 |    },
 98 |    "outputs": [],
 99 |    "source": [
100 |     "%matplotlib inline\n",
101 |     "import matplotlib.pyplot as plt"
102 |    ]
103 |   },
104 |   {
105 |    "cell_type": "code",
106 |    "execution_count": 72,
107 |    "metadata": {
108 |     "collapsed": false
109 |    },
110 |    "outputs": [
111 |     {
112 |      "name": "stdout",
113 |      "output_type": "stream",
114 |      "text": [
115 |       "[0.27134162187576294]\n"
116 |      ]
117 |     },
118 |     {
119 |      "data": {
120 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQQAAAD/CAYAAAAXKqhkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nWCOSk6+uAAAAAElFTkSuQmCC\n",
121 |       "text/plain": [
122 |        "<matplotlib.figure.Figure at 0x145a94438>"
123 |       ]
124 |      },
125 |      "metadata": {},
126 |      "output_type": "display_data"
127 |     }
128 |    ],
129 |    "source": [
130 |     "plt.imshow(image_data[0,:,:,0])\n",
131 |     "print(pr[0])"
132 |    ]
133 |   },
134 |   {
135 |    "cell_type": "code",
136 |    "execution_count": 73,
137 |    "metadata": {
138 |     "collapsed": false
139 |    },
140 |    "outputs": [],
141 |    "source": [
142 |     "g = model.net.graph"
143 |    ]
144 |   },
145 |   {
146 |    "cell_type": "code",
147 |    "execution_count": 74,
148 |    "metadata": {
149 |     "collapsed": false
150 |    },
151 |    "outputs": [
152 |     {
153 |      "data": {
154 |       "text/plain": [
155 |        "['input/X',\n",
156 |        " 'Conv2D/W',\n",
157 |        " 'Conv2D/W/Initializer/random_uniform/shape',\n",
158 |        " 'Conv2D/W/Initializer/random_uniform/min',\n",
159 |        " 'Conv2D/W/Initializer/random_uniform/max',\n",
160 |        " 'Conv2D/W/Initializer/random_uniform/RandomUniform',\n",
161 |        " 'Conv2D/W/Initializer/random_uniform/sub',\n",
162 |        " 'Conv2D/W/Initializer/random_uniform/mul',\n",
163 |        " 'Conv2D/W/Initializer/random_uniform',\n",
164 |        " 'Conv2D/W/Assign',\n",
165 |        " 'Conv2D/W/read',\n",
166 |        " 'Conv2D/b',\n",
167 |        " 'Conv2D/b/Initializer/Const',\n",
168 |        " 'Conv2D/b/Assign',\n",
169 |        " 'Conv2D/b/read',\n",
170 |        " 'Conv2D/Conv2D',\n",
171 |        " 'Conv2D/BiasAdd',\n",
172 |        " 'Conv2D/Tanh',\n",
173 |        " 'FullyConnected/W',\n",
174 |        " 'FullyConnected/W/Initializer/truncated_normal/shape',\n",
175 |        " 'FullyConnected/W/Initializer/truncated_normal/mean',\n",
176 |        " 'FullyConnected/W/Initializer/truncated_normal/stddev',\n",
177 |        " 'FullyConnected/W/Initializer/truncated_normal/TruncatedNormal',\n",
178 |        " 'FullyConnected/W/Initializer/truncated_normal/mul',\n",
179 |        " 'FullyConnected/W/Initializer/truncated_normal',\n",
180 |        " 'FullyConnected/W/Assign',\n",
181 |        " 'FullyConnected/W/read',\n",
182 |        " 'FullyConnected/b',\n",
183 |        " 'FullyConnected/b/Initializer/Const',\n",
184 |        " 'FullyConnected/b/Assign',\n",
185 |        " 'FullyConnected/b/read',\n",
186 |        " 'FullyConnected/Reshape/shape',\n",
187 |        " 'FullyConnected/Reshape',\n",
188 |        " 'FullyConnected/MatMul',\n",
189 |        " 'FullyConnected/BiasAdd',\n",
190 |        " 'target/Y',\n",
191 |        " 'Accuracy/ArgMax/dimension',\n",
192 |        " 'Accuracy/ArgMax',\n",
193 |        " 'Accuracy/ArgMax_1/dimension',\n",
194 |        " 'Accuracy/ArgMax_1',\n",
195 |        " 'Accuracy/Equal',\n",
196 |        " 'Accuracy/Cast',\n",
197 |        " 'Accuracy/Const',\n",
198 |        " 'Accuracy/Mean',\n",
199 |        " 'MeanSquare/sub',\n",
200 |        " 'MeanSquare/Square',\n",
201 |        " 'MeanSquare/Const',\n",
202 |        " 'MeanSquare/Mean',\n",
203 |        " 'Training_step/initial_value',\n",
204 |        " 'Training_step',\n",
205 |        " 'Training_step/Assign',\n",
206 |        " 'Training_step/read',\n",
207 |        " 'is_training',\n",
208 |        " 'is_training/Initializer/Const',\n",
209 |        " 'is_training/Assign',\n",
210 |        " 'is_training/read',\n",
211 |        " 'Assign/value',\n",
212 |        " 'Assign',\n",
213 |        " 'Assign_1/value',\n",
214 |        " 'Assign_1',\n",
215 |        " 'Global_Step/initial_value',\n",
216 |        " 'Global_Step',\n",
217 |        " 'Global_Step/Assign',\n",
218 |        " 'Global_Step/read',\n",
219 |        " 'Add/y',\n",
220 |        " 'Add',\n",
221 |        " 'Assign_2',\n",
222 |        " 'val_loss/initial_value',\n",
223 |        " 'val_loss',\n",
224 |        " 'val_loss/Assign',\n",
225 |        " 'val_loss/read',\n",
226 |        " 'val_acc/initial_value',\n",
227 |        " 'val_acc',\n",
228 |        " 'val_acc/Assign',\n",
229 |        " 'val_acc/read',\n",
230 |        " 'zeros',\n",
231 |        " 'Accuracy/Mean/moving_avg',\n",
232 |        " 'Accuracy/Mean/moving_avg/Assign',\n",
233 |        " 'Accuracy/Mean/moving_avg/read',\n",
234 |        " 'moving_avg/decay',\n",
235 |        " 'moving_avg/add/x',\n",
236 |        " 'moving_avg/add',\n",
237 |        " 'moving_avg/add_1/x',\n",
238 |        " 'moving_avg/add_1',\n",
239 |        " 'moving_avg/truediv',\n",
240 |        " 'moving_avg/Minimum',\n",
241 |        " 'moving_avg/AssignMovingAvg/sub/x',\n",
242 |        " 'moving_avg/AssignMovingAvg/sub',\n",
243 |        " 'moving_avg/AssignMovingAvg/sub_1',\n",
244 |        " 'moving_avg/AssignMovingAvg/mul',\n",
245 |        " 'moving_avg/AssignMovingAvg',\n",
246 |        " 'moving_avg',\n",
247 |        " 'SGD/Total_Loss',\n",
248 |        " 'SGD/zeros',\n",
249 |        " 'SGD/MeanSquare/Mean/moving_avg',\n",
250 |        " 'SGD/MeanSquare/Mean/moving_avg/Assign',\n",
251 |        " 'SGD/MeanSquare/Mean/moving_avg/read',\n",
252 |        " 'SGD/moving_avg/decay',\n",
253 |        " 'SGD/moving_avg/add/x',\n",
254 |        " 'SGD/moving_avg/add',\n",
255 |        " 'SGD/moving_avg/add_1/x',\n",
256 |        " 'SGD/moving_avg/add_1',\n",
257 |        " 'SGD/moving_avg/truediv',\n",
258 |        " 'SGD/moving_avg/Minimum',\n",
259 |        " 'SGD/moving_avg/AssignMovingAvg/sub/x',\n",
260 |        " 'SGD/moving_avg/AssignMovingAvg/sub',\n",
261 |        " 'SGD/moving_avg/AssignMovingAvg/sub_1',\n",
262 |        " 'SGD/moving_avg/AssignMovingAvg/mul',\n",
263 |        " 'SGD/moving_avg/AssignMovingAvg',\n",
264 |        " 'SGD/moving_avg',\n",
265 |        " 'SGD/ScalarSummary/tags',\n",
266 |        " 'SGD/ScalarSummary',\n",
267 |        " 'SGD/ScalarSummary_1/tags',\n",
268 |        " 'SGD/ScalarSummary_1',\n",
269 |        " 'SGD/gradients/Shape',\n",
270 |        " 'SGD/gradients/Const',\n",
271 |        " 'SGD/gradients/Fill',\n",
272 |        " 'SGD/gradients/MeanSquare/Mean_grad/Reshape/shape',\n",
273 |        " 'SGD/gradients/MeanSquare/Mean_grad/Reshape',\n",
274 |        " 'SGD/gradients/MeanSquare/Mean_grad/Shape',\n",
275 |        " 'SGD/gradients/MeanSquare/Mean_grad/Tile',\n",
276 |        " 'SGD/gradients/MeanSquare/Mean_grad/Shape_1',\n",
277 |        " 'SGD/gradients/MeanSquare/Mean_grad/Shape_2',\n",
278 |        " 'SGD/gradients/MeanSquare/Mean_grad/Const',\n",
279 |        " 'SGD/gradients/MeanSquare/Mean_grad/Prod',\n",
280 |        " 'SGD/gradients/MeanSquare/Mean_grad/Const_1',\n",
281 |        " 'SGD/gradients/MeanSquare/Mean_grad/Prod_1',\n",
282 |        " 'SGD/gradients/MeanSquare/Mean_grad/Maximum/y',\n",
283 |        " 'SGD/gradients/MeanSquare/Mean_grad/Maximum',\n",
284 |        " 'SGD/gradients/MeanSquare/Mean_grad/floordiv',\n",
285 |        " 'SGD/gradients/MeanSquare/Mean_grad/Cast',\n",
286 |        " 'SGD/gradients/MeanSquare/Mean_grad/truediv',\n",
287 |        " 'SGD/gradients/MeanSquare/Square_grad/mul/x',\n",
288 |        " 'SGD/gradients/MeanSquare/Square_grad/mul',\n",
289 |        " 'SGD/gradients/MeanSquare/Square_grad/mul_1',\n",
290 |        " 'SGD/gradients/MeanSquare/sub_grad/Shape',\n",
291 |        " 'SGD/gradients/MeanSquare/sub_grad/Shape_1',\n",
292 |        " 'SGD/gradients/MeanSquare/sub_grad/BroadcastGradientArgs',\n",
293 |        " 'SGD/gradients/MeanSquare/sub_grad/Sum',\n",
294 |        " 'SGD/gradients/MeanSquare/sub_grad/Reshape',\n",
295 |        " 'SGD/gradients/MeanSquare/sub_grad/Sum_1',\n",
296 |        " 'SGD/gradients/MeanSquare/sub_grad/Neg',\n",
297 |        " 'SGD/gradients/MeanSquare/sub_grad/Reshape_1',\n",
298 |        " 'SGD/gradients/FullyConnected/BiasAdd_grad/BiasAddGrad',\n",
299 |        " 'SGD/gradients/FullyConnected/MatMul_grad/MatMul',\n",
300 |        " 'SGD/gradients/FullyConnected/MatMul_grad/MatMul_1',\n",
301 |        " 'SGD/gradients/FullyConnected/Reshape_grad/Shape',\n",
302 |        " 'SGD/gradients/FullyConnected/Reshape_grad/Reshape',\n",
303 |        " 'SGD/gradients/Conv2D/Tanh_grad/TanhGrad',\n",
304 |        " 'SGD/gradients/Conv2D/BiasAdd_grad/BiasAddGrad',\n",
305 |        " 'SGD/gradients/Conv2D/Conv2D_grad/Shape',\n",
306 |        " 'SGD/gradients/Conv2D/Conv2D_grad/Conv2DBackpropInput',\n",
307 |        " 'SGD/gradients/Conv2D/Conv2D_grad/Shape_1',\n",
308 |        " 'SGD/gradients/Conv2D/Conv2D_grad/Conv2DBackpropFilter',\n",
309 |        " 'SGD/global_norm/L2Loss',\n",
310 |        " 'SGD/global_norm/L2Loss_1',\n",
311 |        " 'SGD/global_norm/L2Loss_2',\n",
312 |        " 'SGD/global_norm/L2Loss_3',\n",
313 |        " 'SGD/global_norm/pack',\n",
314 |        " 'SGD/global_norm/Const',\n",
315 |        " 'SGD/global_norm/Sum',\n",
316 |        " 'SGD/global_norm/Const_1',\n",
317 |        " 'SGD/global_norm/mul',\n",
318 |        " 'SGD/global_norm/global_norm',\n",
319 |        " 'SGD/clip_by_global_norm/truediv/x',\n",
320 |        " 'SGD/clip_by_global_norm/truediv',\n",
321 |        " 'SGD/clip_by_global_norm/Const',\n",
322 |        " 'SGD/clip_by_global_norm/Minimum',\n",
323 |        " 'SGD/clip_by_global_norm/mul/x',\n",
324 |        " 'SGD/clip_by_global_norm/mul',\n",
325 |        " 'SGD/clip_by_global_norm/mul_1',\n",
326 |        " 'SGD/clip_by_global_norm/SGD/clip_by_global_norm/_0',\n",
327 |        " 'SGD/clip_by_global_norm/mul_2',\n",
328 |        " 'SGD/clip_by_global_norm/SGD/clip_by_global_norm/_1',\n",
329 |        " 'SGD/clip_by_global_norm/mul_3',\n",
330 |        " 'SGD/clip_by_global_norm/SGD/clip_by_global_norm/_2',\n",
331 |        " 'SGD/clip_by_global_norm/mul_4',\n",
332 |        " 'SGD/clip_by_global_norm/SGD/clip_by_global_norm/_3',\n",
333 |        " 'SGD/apply_grad_op_0/learning_rate',\n",
334 |        " 'SGD/apply_grad_op_0/update_Conv2D/W/ApplyGradientDescent',\n",
335 |        " 'SGD/apply_grad_op_0/update_Conv2D/b/ApplyGradientDescent',\n",
336 |        " 'SGD/apply_grad_op_0/update_FullyConnected/W/ApplyGradientDescent',\n",
337 |        " 'SGD/apply_grad_op_0/update_FullyConnected/b/ApplyGradientDescent',\n",
338 |        " 'SGD/apply_grad_op_0/update',\n",
339 |        " 'SGD/apply_grad_op_0/value',\n",
340 |        " 'SGD/apply_grad_op_0',\n",
341 |        " 'SGD/MergeSummary/MergeSummary',\n",
342 |        " 'SGD/train_op_0',\n",
343 |        " 'save/Const',\n",
344 |        " 'save/save/tensor_names',\n",
345 |        " 'save/save/shapes_and_slices',\n",
346 |        " 'save/save',\n",
347 |        " 'save/control_dependency',\n",
348 |        " 'save/restore_slice/tensor_name',\n",
349 |        " 'save/restore_slice/shape_and_slice',\n",
350 |        " 'save/restore_slice',\n",
351 |        " 'save/Assign',\n",
352 |        " 'save/restore_slice_1/tensor_name',\n",
353 |        " 'save/restore_slice_1/shape_and_slice',\n",
354 |        " 'save/restore_slice_1',\n",
355 |        " 'save/Assign_1',\n",
356 |        " 'save/restore_slice_2/tensor_name',\n",
357 |        " 'save/restore_slice_2/shape_and_slice',\n",
358 |        " 'save/restore_slice_2',\n",
359 |        " 'save/Assign_2',\n",
360 |        " 'save/restore_slice_3/tensor_name',\n",
361 |        " 'save/restore_slice_3/shape_and_slice',\n",
362 |        " 'save/restore_slice_3',\n",
363 |        " 'save/Assign_3',\n",
364 |        " 'save/restore_slice_4/tensor_name',\n",
365 |        " 'save/restore_slice_4/shape_and_slice',\n",
366 |        " 'save/restore_slice_4',\n",
367 |        " 'save/Assign_4',\n",
368 |        " 'save/restore_slice_5/tensor_name',\n",
369 |        " 'save/restore_slice_5/shape_and_slice',\n",
370 |        " 'save/restore_slice_5',\n",
371 |        " 'save/Assign_5',\n",
372 |        " 'save/restore_slice_6/tensor_name',\n",
373 |        " 'save/restore_slice_6/shape_and_slice',\n",
374 |        " 'save/restore_slice_6',\n",
375 |        " 'save/Assign_6',\n",
376 |        " 'save/restore_slice_7/tensor_name',\n",
377 |        " 'save/restore_slice_7/shape_and_slice',\n",
378 |        " 'save/restore_slice_7',\n",
379 |        " 'save/Assign_7',\n",
380 |        " 'save/restore_slice_8/tensor_name',\n",
381 |        " 'save/restore_slice_8/shape_and_slice',\n",
382 |        " 'save/restore_slice_8',\n",
383 |        " 'save/Assign_8',\n",
384 |        " 'save/restore_slice_9/tensor_name',\n",
385 |        " 'save/restore_slice_9/shape_and_slice',\n",
386 |        " 'save/restore_slice_9',\n",
387 |        " 'save/Assign_9',\n",
388 |        " 'save/restore_slice_10/tensor_name',\n",
389 |        " 'save/restore_slice_10/shape_and_slice',\n",
390 |        " 'save/restore_slice_10',\n",
391 |        " 'save/Assign_10',\n",
392 |        " 'save/restore_all',\n",
393 |        " 'save_1/Const',\n",
394 |        " 'save_1/save/tensor_names',\n",
395 |        " 'save_1/save/shapes_and_slices',\n",
396 |        " 'save_1/save',\n",
397 |        " 'save_1/control_dependency',\n",
398 |        " 'save_1/restore_slice/tensor_name',\n",
399 |        " 'save_1/restore_slice/shape_and_slice',\n",
400 |        " 'save_1/restore_slice',\n",
401 |        " 'save_1/Assign',\n",
402 |        " 'save_1/restore_slice_1/tensor_name',\n",
403 |        " 'save_1/restore_slice_1/shape_and_slice',\n",
404 |        " 'save_1/restore_slice_1',\n",
405 |        " 'save_1/Assign_1',\n",
406 |        " 'save_1/restore_slice_2/tensor_name',\n",
407 |        " 'save_1/restore_slice_2/shape_and_slice',\n",
408 |        " 'save_1/restore_slice_2',\n",
409 |        " 'save_1/Assign_2',\n",
410 |        " 'save_1/restore_slice_3/tensor_name',\n",
411 |        " 'save_1/restore_slice_3/shape_and_slice',\n",
412 |        " 'save_1/restore_slice_3',\n",
413 |        " 'save_1/Assign_3',\n",
414 |        " 'save_1/restore_slice_4/tensor_name',\n",
415 |        " 'save_1/restore_slice_4/shape_and_slice',\n",
416 |        " 'save_1/restore_slice_4',\n",
417 |        " 'save_1/Assign_4',\n",
418 |        " 'save_1/restore_slice_5/tensor_name',\n",
419 |        " 'save_1/restore_slice_5/shape_and_slice',\n",
420 |        " 'save_1/restore_slice_5',\n",
421 |        " 'save_1/Assign_5',\n",
422 |        " 'save_1/restore_slice_6/tensor_name',\n",
423 |        " 'save_1/restore_slice_6/shape_and_slice',\n",
424 |        " 'save_1/restore_slice_6',\n",
425 |        " 'save_1/Assign_6',\n",
426 |        " 'save_1/restore_slice_7/tensor_name',\n",
427 |        " 'save_1/restore_slice_7/shape_and_slice',\n",
428 |        " 'save_1/restore_slice_7',\n",
429 |        " 'save_1/Assign_7',\n",
430 |        " 'save_1/restore_slice_8/tensor_name',\n",
431 |        " 'save_1/restore_slice_8/shape_and_slice',\n",
432 |        " 'save_1/restore_slice_8',\n",
433 |        " 'save_1/Assign_8',\n",
434 |        " 'save_1/restore_slice_9/tensor_name',\n",
435 |        " 'save_1/restore_slice_9/shape_and_slice',\n",
436 |        " 'save_1/restore_slice_9',\n",
437 |        " 'save_1/Assign_9',\n",
438 |        " 'save_1/restore_slice_10/tensor_name',\n",
439 |        " 'save_1/restore_slice_10/shape_and_slice',\n",
440 |        " 'save_1/restore_slice_10',\n",
441 |        " 'save_1/Assign_10',\n",
442 |        " 'save_1/restore_all',\n",
443 |        " 'init',\n",
444 |        " 'save_2/Const',\n",
445 |        " 'save_2/save/tensor_names',\n",
446 |        " 'save_2/save/shapes_and_slices',\n",
447 |        " 'save_2/save',\n",
448 |        " 'save_2/control_dependency',\n",
449 |        " 'save_2/restore_slice/tensor_name',\n",
450 |        " 'save_2/restore_slice/shape_and_slice',\n",
451 |        " 'save_2/restore_slice',\n",
452 |        " 'save_2/Assign',\n",
453 |        " 'save_2/restore_slice_1/tensor_name',\n",
454 |        " 'save_2/restore_slice_1/shape_and_slice',\n",
455 |        " 'save_2/restore_slice_1',\n",
456 |        " 'save_2/Assign_1',\n",
457 |        " 'save_2/restore_slice_2/tensor_name',\n",
458 |        " 'save_2/restore_slice_2/shape_and_slice',\n",
459 |        " 'save_2/restore_slice_2',\n",
460 |        " 'save_2/Assign_2',\n",
461 |        " 'save_2/restore_slice_3/tensor_name',\n",
462 |        " 'save_2/restore_slice_3/shape_and_slice',\n",
463 |        " 'save_2/restore_slice_3',\n",
464 |        " 'save_2/Assign_3',\n",
465 |        " 'save_2/restore_slice_4/tensor_name',\n",
466 |        " 'save_2/restore_slice_4/shape_and_slice',\n",
467 |        " 'save_2/restore_slice_4',\n",
468 |        " 'save_2/Assign_4',\n",
469 |        " 'save_2/restore_slice_5/tensor_name',\n",
470 |        " 'save_2/restore_slice_5/shape_and_slice',\n",
471 |        " 'save_2/restore_slice_5',\n",
472 |        " 'save_2/Assign_5',\n",
473 |        " 'save_2/restore_slice_6/tensor_name',\n",
474 |        " 'save_2/restore_slice_6/shape_and_slice',\n",
475 |        " 'save_2/restore_slice_6',\n",
476 |        " 'save_2/Assign_6',\n",
477 |        " 'save_2/restore_slice_7/tensor_name',\n",
478 |        " 'save_2/restore_slice_7/shape_and_slice',\n",
479 |        " 'save_2/restore_slice_7',\n",
480 |        " 'save_2/Assign_7',\n",
481 |        " 'save_2/restore_slice_8/tensor_name',\n",
482 |        " 'save_2/restore_slice_8/shape_and_slice',\n",
483 |        " 'save_2/restore_slice_8',\n",
484 |        " 'save_2/Assign_8',\n",
485 |        " 'save_2/restore_slice_9/tensor_name',\n",
486 |        " 'save_2/restore_slice_9/shape_and_slice',\n",
487 |        " 'save_2/restore_slice_9',\n",
488 |        " 'save_2/Assign_9',\n",
489 |        " 'save_2/restore_slice_10/tensor_name',\n",
490 |        " 'save_2/restore_slice_10/shape_and_slice',\n",
491 |        " 'save_2/restore_slice_10',\n",
492 |        " 'save_2/Assign_10',\n",
493 |        " 'save_2/restore_all',\n",
494 |        " 'init_1',\n",
495 |        " 'save_3/Const',\n",
496 |        " 'save_3/save/tensor_names',\n",
497 |        " 'save_3/save/shapes_and_slices',\n",
498 |        " 'save_3/save',\n",
499 |        " 'save_3/control_dependency',\n",
500 |        " 'save_3/restore_slice/tensor_name',\n",
501 |        " 'save_3/restore_slice/shape_and_slice',\n",
502 |        " 'save_3/restore_slice',\n",
503 |        " 'save_3/Assign',\n",
504 |        " 'save_3/restore_slice_1/tensor_name',\n",
505 |        " 'save_3/restore_slice_1/shape_and_slice',\n",
506 |        " 'save_3/restore_slice_1',\n",
507 |        " 'save_3/Assign_1',\n",
508 |        " 'save_3/restore_slice_2/tensor_name',\n",
509 |        " 'save_3/restore_slice_2/shape_and_slice',\n",
510 |        " 'save_3/restore_slice_2',\n",
511 |        " 'save_3/Assign_2',\n",
512 |        " 'save_3/restore_slice_3/tensor_name',\n",
513 |        " 'save_3/restore_slice_3/shape_and_slice',\n",
514 |        " 'save_3/restore_slice_3',\n",
515 |        " 'save_3/Assign_3',\n",
516 |        " 'save_3/restore_slice_4/tensor_name',\n",
517 |        " 'save_3/restore_slice_4/shape_and_slice',\n",
518 |        " 'save_3/restore_slice_4',\n",
519 |        " 'save_3/Assign_4',\n",
520 |        " 'save_3/restore_slice_5/tensor_name',\n",
521 |        " 'save_3/restore_slice_5/shape_and_slice',\n",
522 |        " 'save_3/restore_slice_5',\n",
523 |        " 'save_3/Assign_5',\n",
524 |        " 'save_3/restore_slice_6/tensor_name',\n",
525 |        " 'save_3/restore_slice_6/shape_and_slice',\n",
526 |        " 'save_3/restore_slice_6',\n",
527 |        " 'save_3/Assign_6',\n",
528 |        " 'save_3/restore_slice_7/tensor_name',\n",
529 |        " 'save_3/restore_slice_7/shape_and_slice',\n",
530 |        " 'save_3/restore_slice_7',\n",
531 |        " 'save_3/Assign_7',\n",
532 |        " 'save_3/restore_slice_8/tensor_name',\n",
533 |        " 'save_3/restore_slice_8/shape_and_slice',\n",
534 |        " 'save_3/restore_slice_8',\n",
535 |        " 'save_3/Assign_8',\n",
536 |        " 'save_3/restore_slice_9/tensor_name',\n",
537 |        " 'save_3/restore_slice_9/shape_and_slice',\n",
538 |        " 'save_3/restore_slice_9',\n",
539 |        " 'save_3/Assign_9',\n",
540 |        " 'save_3/restore_slice_10/tensor_name',\n",
541 |        " 'save_3/restore_slice_10/shape_and_slice',\n",
542 |        " 'save_3/restore_slice_10',\n",
543 |        " 'save_3/Assign_10',\n",
544 |        " 'save_3/restore_all',\n",
545 |        " 'init_2',\n",
546 |        " 'save_4/Const',\n",
547 |        " 'save_4/save/tensor_names',\n",
548 |        " 'save_4/save/shapes_and_slices',\n",
549 |        " 'save_4/save',\n",
550 |        " 'save_4/control_dependency',\n",
551 |        " 'save_4/restore_slice/tensor_name',\n",
552 |        " 'save_4/restore_slice/shape_and_slice',\n",
553 |        " 'save_4/restore_slice',\n",
554 |        " 'save_4/Assign',\n",
555 |        " 'save_4/restore_slice_1/tensor_name',\n",
556 |        " 'save_4/restore_slice_1/shape_and_slice',\n",
557 |        " 'save_4/restore_slice_1',\n",
558 |        " 'save_4/Assign_1',\n",
559 |        " 'save_4/restore_slice_2/tensor_name',\n",
560 |        " 'save_4/restore_slice_2/shape_and_slice',\n",
561 |        " 'save_4/restore_slice_2',\n",
562 |        " 'save_4/Assign_2',\n",
563 |        " 'save_4/restore_slice_3/tensor_name',\n",
564 |        " 'save_4/restore_slice_3/shape_and_slice',\n",
565 |        " 'save_4/restore_slice_3',\n",
566 |        " 'save_4/Assign_3',\n",
567 |        " 'save_4/restore_slice_4/tensor_name',\n",
568 |        " 'save_4/restore_slice_4/shape_and_slice',\n",
569 |        " 'save_4/restore_slice_4',\n",
570 |        " 'save_4/Assign_4',\n",
571 |        " 'save_4/restore_slice_5/tensor_name',\n",
572 |        " 'save_4/restore_slice_5/shape_and_slice',\n",
573 |        " 'save_4/restore_slice_5',\n",
574 |        " 'save_4/Assign_5',\n",
575 |        " 'save_4/restore_slice_6/tensor_name',\n",
576 |        " 'save_4/restore_slice_6/shape_and_slice',\n",
577 |        " 'save_4/restore_slice_6',\n",
578 |        " 'save_4/Assign_6',\n",
579 |        " 'save_4/restore_slice_7/tensor_name',\n",
580 |        " 'save_4/restore_slice_7/shape_and_slice',\n",
581 |        " 'save_4/restore_slice_7',\n",
582 |        " 'save_4/Assign_7',\n",
583 |        " 'save_4/restore_slice_8/tensor_name',\n",
584 |        " 'save_4/restore_slice_8/shape_and_slice',\n",
585 |        " 'save_4/restore_slice_8',\n",
586 |        " 'save_4/Assign_8',\n",
587 |        " 'save_4/restore_slice_9/tensor_name',\n",
588 |        " 'save_4/restore_slice_9/shape_and_slice',\n",
589 |        " 'save_4/restore_slice_9',\n",
590 |        " 'save_4/Assign_9',\n",
591 |        " 'save_4/restore_slice_10/tensor_name',\n",
592 |        " 'save_4/restore_slice_10/shape_and_slice',\n",
593 |        " 'save_4/restore_slice_10',\n",
594 |        " 'save_4/Assign_10',\n",
595 |        " 'save_4/restore_all',\n",
596 |        " 'init_3',\n",
597 |        " 'save_5/Const',\n",
598 |        " 'save_5/save/tensor_names',\n",
599 |        " 'save_5/save/shapes_and_slices',\n",
600 |        " 'save_5/save',\n",
601 |        " 'save_5/control_dependency',\n",
602 |        " 'save_5/restore_slice/tensor_name',\n",
603 |        " 'save_5/restore_slice/shape_and_slice',\n",
604 |        " 'save_5/restore_slice',\n",
605 |        " 'save_5/Assign',\n",
606 |        " 'save_5/restore_slice_1/tensor_name',\n",
607 |        " 'save_5/restore_slice_1/shape_and_slice',\n",
608 |        " 'save_5/restore_slice_1',\n",
609 |        " 'save_5/Assign_1',\n",
610 |        " 'save_5/restore_slice_2/tensor_name',\n",
611 |        " 'save_5/restore_slice_2/shape_and_slice',\n",
612 |        " 'save_5/restore_slice_2',\n",
613 |        " 'save_5/Assign_2',\n",
614 |        " 'save_5/restore_slice_3/tensor_name',\n",
615 |        " 'save_5/restore_slice_3/shape_and_slice',\n",
616 |        " 'save_5/restore_slice_3',\n",
617 |        " 'save_5/Assign_3',\n",
618 |        " 'save_5/restore_slice_4/tensor_name',\n",
619 |        " 'save_5/restore_slice_4/shape_and_slice',\n",
620 |        " 'save_5/restore_slice_4',\n",
621 |        " 'save_5/Assign_4',\n",
622 |        " 'save_5/restore_slice_5/tensor_name',\n",
623 |        " 'save_5/restore_slice_5/shape_and_slice',\n",
624 |        " 'save_5/restore_slice_5',\n",
625 |        " 'save_5/Assign_5',\n",
626 |        " 'save_5/restore_slice_6/tensor_name',\n",
627 |        " 'save_5/restore_slice_6/shape_and_slice',\n",
628 |        " 'save_5/restore_slice_6',\n",
629 |        " 'save_5/Assign_6',\n",
630 |        " 'save_5/restore_slice_7/tensor_name',\n",
631 |        " 'save_5/restore_slice_7/shape_and_slice',\n",
632 |        " 'save_5/restore_slice_7',\n",
633 |        " 'save_5/Assign_7',\n",
634 |        " 'save_5/restore_slice_8/tensor_name',\n",
635 |        " 'save_5/restore_slice_8/shape_and_slice',\n",
636 |        " 'save_5/restore_slice_8',\n",
637 |        " 'save_5/Assign_8',\n",
638 |        " 'save_5/restore_slice_9/tensor_name',\n",
639 |        " 'save_5/restore_slice_9/shape_and_slice',\n",
640 |        " 'save_5/restore_slice_9',\n",
641 |        " 'save_5/Assign_9',\n",
642 |        " 'save_5/restore_slice_10/tensor_name',\n",
643 |        " 'save_5/restore_slice_10/shape_and_slice',\n",
644 |        " 'save_5/restore_slice_10',\n",
645 |        " 'save_5/Assign_10',\n",
646 |        " 'save_5/restore_all',\n",
647 |        " 'init_4',\n",
648 |        " 'save_6/Const',\n",
649 |        " 'save_6/save/tensor_names',\n",
650 |        " 'save_6/save/shapes_and_slices',\n",
651 |        " 'save_6/save',\n",
652 |        " 'save_6/control_dependency',\n",
653 |        " 'save_6/restore_slice/tensor_name',\n",
654 |        " 'save_6/restore_slice/shape_and_slice',\n",
655 |        " 'save_6/restore_slice',\n",
656 |        " 'save_6/Assign',\n",
657 |        " 'save_6/restore_slice_1/tensor_name',\n",
658 |        " 'save_6/restore_slice_1/shape_and_slice',\n",
659 |        " 'save_6/restore_slice_1',\n",
660 |        " 'save_6/Assign_1',\n",
661 |        " 'save_6/restore_slice_2/tensor_name',\n",
662 |        " 'save_6/restore_slice_2/shape_and_slice',\n",
663 |        " 'save_6/restore_slice_2',\n",
664 |        " 'save_6/Assign_2',\n",
665 |        " 'save_6/restore_slice_3/tensor_name',\n",
666 |        " 'save_6/restore_slice_3/shape_and_slice',\n",
667 |        " 'save_6/restore_slice_3',\n",
668 |        " 'save_6/Assign_3',\n",
669 |        " 'save_6/restore_slice_4/tensor_name',\n",
670 |        " 'save_6/restore_slice_4/shape_and_slice',\n",
671 |        " 'save_6/restore_slice_4',\n",
672 |        " 'save_6/Assign_4',\n",
673 |        " 'save_6/restore_slice_5/tensor_name',\n",
674 |        " 'save_6/restore_slice_5/shape_and_slice',\n",
675 |        " 'save_6/restore_slice_5',\n",
676 |        " 'save_6/Assign_5',\n",
677 |        " 'save_6/restore_slice_6/tensor_name',\n",
678 |        " 'save_6/restore_slice_6/shape_and_slice',\n",
679 |        " 'save_6/restore_slice_6',\n",
680 |        " 'save_6/Assign_6',\n",
681 |        " 'save_6/restore_slice_7/tensor_name',\n",
682 |        " 'save_6/restore_slice_7/shape_and_slice',\n",
683 |        " 'save_6/restore_slice_7',\n",
684 |        " 'save_6/Assign_7',\n",
685 |        " 'save_6/restore_slice_8/tensor_name',\n",
686 |        " 'save_6/restore_slice_8/shape_and_slice',\n",
687 |        " 'save_6/restore_slice_8',\n",
688 |        " 'save_6/Assign_8',\n",
689 |        " 'save_6/restore_slice_9/tensor_name',\n",
690 |        " 'save_6/restore_slice_9/shape_and_slice',\n",
691 |        " 'save_6/restore_slice_9',\n",
692 |        " 'save_6/Assign_9',\n",
693 |        " 'save_6/restore_slice_10/tensor_name',\n",
694 |        " 'save_6/restore_slice_10/shape_and_slice',\n",
695 |        " 'save_6/restore_slice_10',\n",
696 |        " 'save_6/Assign_10',\n",
697 |        " 'save_6/restore_all',\n",
698 |        " 'init_5',\n",
699 |        " 'init_6',\n",
700 |        " 'save_7/Const',\n",
701 |        " 'save_7/save/tensor_names',\n",
702 |        " 'save_7/save/shapes_and_slices',\n",
703 |        " 'save_7/save',\n",
704 |        " 'save_7/control_dependency',\n",
705 |        " 'save_7/restore_slice/tensor_name',\n",
706 |        " 'save_7/restore_slice/shape_and_slice',\n",
707 |        " 'save_7/restore_slice',\n",
708 |        " 'save_7/Assign',\n",
709 |        " 'save_7/restore_slice_1/tensor_name',\n",
710 |        " 'save_7/restore_slice_1/shape_and_slice',\n",
711 |        " 'save_7/restore_slice_1',\n",
712 |        " 'save_7/Assign_1',\n",
713 |        " 'save_7/restore_slice_2/tensor_name',\n",
714 |        " 'save_7/restore_slice_2/shape_and_slice',\n",
715 |        " 'save_7/restore_slice_2',\n",
716 |        " 'save_7/Assign_2',\n",
717 |        " 'save_7/restore_slice_3/tensor_name',\n",
718 |        " 'save_7/restore_slice_3/shape_and_slice',\n",
719 |        " 'save_7/restore_slice_3',\n",
720 |        " 'save_7/Assign_3',\n",
721 |        " 'save_7/restore_slice_4/tensor_name',\n",
722 |        " 'save_7/restore_slice_4/shape_and_slice',\n",
723 |        " 'save_7/restore_slice_4',\n",
724 |        " 'save_7/Assign_4',\n",
725 |        " 'save_7/restore_slice_5/tensor_name',\n",
726 |        " 'save_7/restore_slice_5/shape_and_slice',\n",
727 |        " 'save_7/restore_slice_5',\n",
728 |        " 'save_7/Assign_5',\n",
729 |        " 'save_7/restore_slice_6/tensor_name',\n",
730 |        " 'save_7/restore_slice_6/shape_and_slice',\n",
731 |        " 'save_7/restore_slice_6',\n",
732 |        " 'save_7/Assign_6',\n",
733 |        " 'save_7/restore_slice_7/tensor_name',\n",
734 |        " 'save_7/restore_slice_7/shape_and_slice',\n",
735 |        " 'save_7/restore_slice_7',\n",
736 |        " 'save_7/Assign_7',\n",
737 |        " 'save_7/restore_slice_8/tensor_name',\n",
738 |        " 'save_7/restore_slice_8/shape_and_slice',\n",
739 |        " 'save_7/restore_slice_8',\n",
740 |        " 'save_7/Assign_8',\n",
741 |        " 'save_7/restore_slice_9/tensor_name',\n",
742 |        " 'save_7/restore_slice_9/shape_and_slice',\n",
743 |        " 'save_7/restore_slice_9',\n",
744 |        " 'save_7/Assign_9',\n",
745 |        " 'save_7/restore_slice_10/tensor_name',\n",
746 |        " 'save_7/restore_slice_10/shape_and_slice',\n",
747 |        " 'save_7/restore_slice_10',\n",
748 |        " 'save_7/Assign_10',\n",
749 |        " 'save_7/restore_all']"
750 |       ]
751 |      },
752 |      "execution_count": 74,
753 |      "metadata": {},
754 |      "output_type": "execute_result"
755 |     }
756 |    ],
757 |    "source": [
758 |     "[op.name for op in g.get_operations()]"
759 |    ]
760 |   },
761 |   {
762 |    "cell_type": "code",
763 |    "execution_count": 75,
764 |    "metadata": {
765 |     "collapsed": false
766 |    },
767 |    "outputs": [],
768 |    "source": [
769 |     "conv2_tensor = g.get_tensor_by_name('Conv2D/Tanh:0')\n",
770 |     "shape = conv2_tensor.get_shape()\n",
771 |     "\n",
772 |     "fc_W_tensor = g.get_tensor_by_name('FullyConnected/W:0')\n",
773 |     "fc_W = fc_W_tensor.eval(session=model.session)\n",
774 |     "fc_W = fc_W.reshape(*shape[1:])"
775 |    ]
776 |   },
777 |   {
778 |    "cell_type": "code",
779 |    "execution_count": 76,
780 |    "metadata": {
781 |     "collapsed": false
782 |    },
783 |    "outputs": [],
784 |    "source": [
785 |     "%matplotlib inline\n",
786 |     "import matplotlib.pyplot as plt"
787 |    ]
788 |   },
789 |   {
790 |    "cell_type": "code",
791 |    "execution_count": 77,
792 |    "metadata": {
793 |     "collapsed": false
794 |    },
795 |    "outputs": [
796 |     {
797 |      "data": {
798 |       "text/plain": [
799 |        "<matplotlib.image.AxesImage at 0x14605a908>"
800 |       ]
801 |      },
802 |      "execution_count": 77,
803 |      "metadata": {},
804 |      "output_type": "execute_result"
805 |     },
806 |     {
807 |      "data": {
808 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP4AAAD7CAYAAABKWyniAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nHISb+A4OOQg38R0cchBu4js45CD+XxIKu/JH9qkuAAAAAElFTkSuQmCC\n",
809 |       "text/plain": [
810 |        "<matplotlib.figure.Figure at 0x144f9e2b0>"
811 |       ]
812 |      },
813 |      "metadata": {},
814 |      "output_type": "display_data"
815 |     }
816 |    ],
817 |    "source": [
818 |     "plt.matshow(fc_W[:,:,0])"
819 |    ]
820 |   },
821 |   {
822 |    "cell_type": "code",
823 |    "execution_count": null,
824 |    "metadata": {
825 |     "collapsed": true
826 |    },
827 |    "outputs": [],
828 |    "source": []
829 |   }
830 |  ],
831 |  "metadata": {
832 |   "kernelspec": {
833 |    "display_name": "Python [py3]",
834 |    "language": "python",
835 |    "name": "Python [py3]"
836 |   },
837 |   "language_info": {
838 |    "codemirror_mode": {
839 |     "name": "ipython",
840 |     "version": 3
841 |    },
842 |    "file_extension": ".py",
843 |    "mimetype": "text/x-python",
844 |    "name": "python",
845 |    "nbconvert_exporter": "python",
846 |    "pygments_lexer": "ipython3",
847 |    "version": "3.5.2"
848 |   }
849 |  },
850 |  "nbformat": 4,
851 |  "nbformat_minor": 0
852 | }
853 | 


--------------------------------------------------------------------------------
/src/ai/explore_train.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 3,
  6 |    "metadata": {
  7 |     "collapsed": true
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "import numpy as np"
 12 |    ]
 13 |   },
 14 |   {
 15 |    "cell_type": "code",
 16 |    "execution_count": 4,
 17 |    "metadata": {
 18 |     "collapsed": true
 19 |    },
 20 |    "outputs": [],
 21 |    "source": [
 22 |     "import tensorflow as tf"
 23 |    ]
 24 |   },
 25 |   {
 26 |    "cell_type": "code",
 27 |    "execution_count": 5,
 28 |    "metadata": {
 29 |     "collapsed": true
 30 |    },
 31 |    "outputs": [],
 32 |    "source": [
 33 |     "import tflearn"
 34 |    ]
 35 |   },
 36 |   {
 37 |    "cell_type": "code",
 38 |    "execution_count": 7,
 39 |    "metadata": {
 40 |     "collapsed": false
 41 |    },
 42 |    "outputs": [],
 43 |    "source": [
 44 |     "%reload_ext autoreload\n",
 45 |     "%aimport generate_data"
 46 |    ]
 47 |   },
 48 |   {
 49 |    "cell_type": "code",
 50 |    "execution_count": 45,
 51 |    "metadata": {
 52 |     "collapsed": false
 53 |    },
 54 |    "outputs": [],
 55 |    "source": [
 56 |     "input_size = 128\n",
 57 |     "minibatch_size = 10000\n",
 58 |     "image_data, label_data = generate_data.generate_batch(height=input_size, width=input_size, minibatch_size=minibatch_size)"
 59 |    ]
 60 |   },
 61 |   {
 62 |    "cell_type": "code",
 63 |    "execution_count": 26,
 64 |    "metadata": {
 65 |     "collapsed": true
 66 |    },
 67 |    "outputs": [],
 68 |    "source": [
 69 |     "from tflearn.layers.core import input_data, dropout, fully_connected\n",
 70 |     "from tflearn.layers.conv import conv_2d, max_pool_2d\n",
 71 |     "from tflearn.layers.normalization import local_response_normalization\n",
 72 |     "from tflearn.layers.estimator import regression"
 73 |    ]
 74 |   },
 75 |   {
 76 |    "cell_type": "code",
 77 |    "execution_count": 49,
 78 |    "metadata": {
 79 |     "collapsed": true
 80 |    },
 81 |    "outputs": [],
 82 |    "source": [
 83 |     "tf.reset_default_graph()"
 84 |    ]
 85 |   },
 86 |   {
 87 |    "cell_type": "code",
 88 |    "execution_count": 50,
 89 |    "metadata": {
 90 |     "collapsed": false
 91 |    },
 92 |    "outputs": [],
 93 |    "source": [
 94 |     "# Building convolutional network\n",
 95 |     "network = input_data(shape=[None, 128, 128, 1], name='input')\n",
 96 |     "network = conv_2d(network, nb_filter=2, filter_size=5, strides=1, activation='tanh')\n",
 97 |     "network = fully_connected(network, 1, activation='linear')\n",
 98 |     "network = regression(network, optimizer='adam', learning_rate=0.001,\n",
 99 |     "                     loss='mean_square', name='target')"
100 |    ]
101 |   },
102 |   {
103 |    "cell_type": "code",
104 |    "execution_count": 51,
105 |    "metadata": {
106 |     "collapsed": false
107 |    },
108 |    "outputs": [
109 |     {
110 |      "name": "stdout",
111 |      "output_type": "stream",
112 |      "text": [
113 |       "Training Step: 138  | total loss: \u001b[1m\u001b[32m0.04324\u001b[0m\u001b[0m\n",
114 |       "\u001b[2K\r",
115 |       "| Adam | epoch: 000 | loss: 0.04324 - acc: 1.0000 -- iter: 08832/10000\n"
116 |      ]
117 |     },
118 |     {
119 |      "ename": "KeyboardInterrupt",
120 |      "evalue": "",
121 |      "output_type": "error",
122 |      "traceback": [
123 |       "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
124 |       "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
125 |       "\u001b[0;32m<ipython-input-51-e3dab6e6869f>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0mmodel\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtflearn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mDNN\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnetwork\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtensorboard_verbose\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      6\u001b[0m model.fit({'input': X}, {'target': Y}, n_epoch=100,\n\u001b[0;32m----> 7\u001b[0;31m            snapshot_step=100, show_metric=True, run_id='road_model1')\n\u001b[0m",
126 |       "\u001b[0;32m/Users/andy/anaconda/envs/py3/lib/python3.5/site-packages/tflearn/models/dnn.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, X_inputs, Y_targets, n_epoch, validation_set, show_metric, batch_size, shuffle, snapshot_epoch, snapshot_step, excl_trainops, run_id)\u001b[0m\n\u001b[1;32m    186\u001b[0m                          \u001b[0mdaug_dict\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mdaug_dict\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    187\u001b[0m                          \u001b[0mexcl_trainops\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mexcl_trainops\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 188\u001b[0;31m                          run_id=run_id)\n\u001b[0m\u001b[1;32m    189\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    190\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mpredict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mX\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
127 |       "\u001b[0;32m/Users/andy/anaconda/envs/py3/lib/python3.5/site-packages/tflearn/helpers/trainer.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, feed_dicts, n_epoch, val_feed_dicts, show_metric, snapshot_step, snapshot_epoch, shuffle_all, dprep_dict, daug_dict, excl_trainops, run_id)\u001b[0m\n\u001b[1;32m    275\u001b[0m                                                        \u001b[0msnapshot_epoch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    276\u001b[0m                                                        \u001b[0msnapshot_step\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 277\u001b[0;31m                                                        show_metric)\n\u001b[0m\u001b[1;32m    278\u001b[0m                             \u001b[0mglobal_loss\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0mtrain_op\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mloss_value\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    279\u001b[0m                             \u001b[0;32mif\u001b[0m \u001b[0mtrain_op\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0macc_value\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mglobal_acc\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
128 |       "\u001b[0;32m/Users/andy/anaconda/envs/py3/lib/python3.5/site-packages/tflearn/helpers/trainer.py\u001b[0m in \u001b[0;36m_train\u001b[0;34m(self, training_step, snapshot_epoch, snapshot_step, show_metric)\u001b[0m\n\u001b[1;32m    682\u001b[0m         \u001b[0mtflearn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mis_training\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msession\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msession\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    683\u001b[0m         _, train_summ_str = self.session.run([self.train, self.summ_op],\n\u001b[0;32m--> 684\u001b[0;31m                                              feed_batch)\n\u001b[0m\u001b[1;32m    685\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    686\u001b[0m         \u001b[0;31m# Retrieve loss value from summary string\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
129 |       "\u001b[0;32m/Users/andy/anaconda/envs/py3/lib/python3.5/site-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m    370\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    371\u001b[0m       result = self._run(None, fetches, feed_dict, options_ptr,\n\u001b[0;32m--> 372\u001b[0;31m                          run_metadata_ptr)\n\u001b[0m\u001b[1;32m    373\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    374\u001b[0m         \u001b[0mproto_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf_session\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrun_metadata_ptr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
130 |       "\u001b[0;32m/Users/andy/anaconda/envs/py3/lib/python3.5/site-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_run\u001b[0;34m(self, handle, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m    634\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    635\u001b[0m       results = self._do_run(handle, target_list, unique_fetches,\n\u001b[0;32m--> 636\u001b[0;31m                              feed_dict_string, options, run_metadata)\n\u001b[0m\u001b[1;32m    637\u001b[0m     \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    638\u001b[0m       \u001b[0;31m# The movers are no longer used. Delete them.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
131 |       "\u001b[0;32m/Users/andy/anaconda/envs/py3/lib/python3.5/site-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_do_run\u001b[0;34m(self, handle, target_list, fetch_list, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m    706\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mhandle\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    707\u001b[0m       return self._do_call(_run_fn, self._session, feed_dict, fetch_list,\n\u001b[0;32m--> 708\u001b[0;31m                            target_list, options, run_metadata)\n\u001b[0m\u001b[1;32m    709\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    710\u001b[0m       return self._do_call(_prun_fn, self._session, handle, feed_dict,\n",
132 |       "\u001b[0;32m/Users/andy/anaconda/envs/py3/lib/python3.5/site-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_do_call\u001b[0;34m(self, fn, *args)\u001b[0m\n\u001b[1;32m    713\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    714\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 715\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    716\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mOpError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    717\u001b[0m       \u001b[0mmessage\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompat\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mas_text\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmessage\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
133 |       "\u001b[0;32m/Users/andy/anaconda/envs/py3/lib/python3.5/site-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_run_fn\u001b[0;34m(session, feed_dict, fetch_list, target_list, options, run_metadata)\u001b[0m\n\u001b[1;32m    695\u001b[0m         return tf_session.TF_Run(session, options,\n\u001b[1;32m    696\u001b[0m                                  \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget_list\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 697\u001b[0;31m                                  status, run_metadata)\n\u001b[0m\u001b[1;32m    698\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    699\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_prun_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msession\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
134 |       "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
135 |      ]
136 |     }
137 |    ],
138 |    "source": [
139 |     "X = image_data\n",
140 |     "Y = label_data[:,np.newaxis]\n",
141 |     "\n",
142 |     "# Train\n",
143 |     "model = tflearn.DNN(network, tensorboard_verbose=0)\n",
144 |     "model.fit({'input': X}, {'target': Y}, n_epoch=100,\n",
145 |     "           snapshot_step=100, show_metric=True, run_id='road_model1')"
146 |    ]
147 |   },
148 |   {
149 |    "cell_type": "code",
150 |    "execution_count": 52,
151 |    "metadata": {
152 |     "collapsed": true
153 |    },
154 |    "outputs": [],
155 |    "source": [
156 |     "? model.fit"
157 |    ]
158 |   },
159 |   {
160 |    "cell_type": "code",
161 |    "execution_count": null,
162 |    "metadata": {
163 |     "collapsed": true
164 |    },
165 |    "outputs": [],
166 |    "source": []
167 |   }
168 |  ],
169 |  "metadata": {
170 |   "anaconda-cloud": {},
171 |   "kernelspec": {
172 |    "display_name": "Python [py3]",
173 |    "language": "python",
174 |    "name": "Python [py3]"
175 |   },
176 |   "language_info": {
177 |    "codemirror_mode": {
178 |     "name": "ipython",
179 |     "version": 3
180 |    },
181 |    "file_extension": ".py",
182 |    "mimetype": "text/x-python",
183 |    "name": "python",
184 |    "nbconvert_exporter": "python",
185 |    "pygments_lexer": "ipython3",
186 |    "version": "3.5.2"
187 |   }
188 |  },
189 |  "nbformat": 4,
190 |  "nbformat_minor": 0
191 | }
192 | 


--------------------------------------------------------------------------------
/src/ai/generate_data.py:
--------------------------------------------------------------------------------
 1 | import math
 2 | import numpy as np
 3 | import skimage
 4 | import skimage.draw
 5 | 
 6 | def create_road(height=128, width=128, offset=0.0):
 7 |   image = np.zeros((height, width), dtype=np.uint8)
 8 |   left_edge  = width / 3
 9 |   right_edge = width - left_edge
10 | 
11 |   if (math.fabs(offset) > left_edge):
12 |     offset = math.copysign(left_edge, offset)
13 | 
14 |   rows, columns = skimage.draw.polygon(  \
15 |     [0, height, height, 0],              \
16 |     [left_edge + offset, left_edge, right_edge, right_edge + offset])
17 |   image[rows, columns] = 1
18 |   image = image[:, :, np.newaxis]
19 |   return image
20 | 
21 | 
22 | def random_road(height=128, width=128):
23 |   offset = (np.random.rand() * 256 - 128) / 3
24 |   image = create_road(height, width, offset)
25 |   return image, offset
26 | 
27 | 
28 | def normal_distribution(images):
29 |   np.add(images, -.5, out=images)
30 |   np.multiply(images, 4, out=images)
31 | 
32 | 
33 | def generate_batch(height=128, width=128, minibatch_size=10):
34 |   input_data = np.zeros([minibatch_size, height, width, 1], dtype=np.float32)
35 |   label_data = np.zeros([minibatch_size], dtype=np.float32)
36 | 
37 |   for i in range(minibatch_size):
38 |     image, offset = random_road(height, width)
39 |     input_data[i, :, :, :] = image
40 |     label_data[i] = offset / width
41 | 
42 |   normal_distribution(input_data)
43 | 
44 |   return input_data, label_data
45 | 


--------------------------------------------------------------------------------
/src/ai/run.py:
--------------------------------------------------------------------------------
 1 | import tensorflow as tf
 2 | import tflearn
 3 | 
 4 | import train
 5 | import generate_data
 6 | 
 7 | tf.reset_default_graph()
 8 | model = train.build_model()
 9 | 
10 | model.load('checkpoints/road_model1-72')
11 | 
12 | input_size = 128
13 | data_size = 1
14 | 
15 | image_data, label_data = \
16 |     generate_data.generate_batch(
17 |         height=input_size,
18 |         width=input_size,
19 |         minibatch_size=data_size)
20 | 
21 | pr = model.predict(image_data)
22 | 
23 | import matplotlib.pyplot as plt
24 | 
25 | print(pr[0])
26 | plt.imshow(image_data[0,:,:,0], interpolation='nearest')
27 | plt.colorbar()
28 | plt.show()
29 | 


--------------------------------------------------------------------------------
/src/ai/train.py:
--------------------------------------------------------------------------------
 1 | import os
 2 | 
 3 | import numpy as np
 4 | from numpy import newaxis
 5 | import tensorflow as tf
 6 | 
 7 | import tflearn
 8 | from tflearn.layers.core import input_data, fully_connected
 9 | from tflearn.layers.conv import conv_2d
10 | from tflearn.layers.estimator import regression
11 | 
12 | import generate_data
13 | 
14 | learning_rate = 0.001
15 | num_iterations = 1000
16 | 
17 | def build_model():
18 | 	init = tf.truncated_normal_initializer(stddev=1e-4)
19 | 
20 | 	network = input_data(shape=[None, 128, 128, 1], name='input')
21 | 	network = conv_2d(network, nb_filter=2, filter_size=5, strides=2, activation='tanh', weights_init=init)
22 | 	network = fully_connected(network, 1, activation='tanh', weights_init=init)
23 | 	network = regression(network, optimizer='sgd', learning_rate=learning_rate,
24 | 						 loss='mean_square', name='target')
25 | 
26 | 	model = tflearn.DNN(network, tensorboard_verbose=0, checkpoint_path='checkpoints/road_model1')
27 | 	return model
28 | 
29 | 
30 | if __name__ == '__main__':
31 | 	input_size = 128
32 | 	minibatch_size = 16
33 | 	batch_size = 3200
34 | 
35 | 	checkpoint_path = 'checkpoints'
36 | 	if not os.path.exists(checkpoint_path):
37 | 		os.makedirs(checkpoint_path)
38 | 
39 | 	model = build_model()
40 | 
41 | 	# Train
42 | 	for i in range(num_iterations):
43 | 		image_data, label_data = \
44 | 			generate_data.generate_batch(
45 | 				height=input_size,
46 | 				width=input_size,
47 | 				minibatch_size=batch_size)
48 | 
49 | 		X = image_data
50 | 		Y = label_data[:,newaxis]
51 | 
52 | 		model.fit({'input': X}, {'target': Y},
53 | 				   n_epoch=1,
54 | 				   batch_size=minibatch_size,
55 | 				   snapshot_epoch=True, show_metric=True, run_id='road_model1')
56 | 


--------------------------------------------------------------------------------
/src/ai/video.py:
--------------------------------------------------------------------------------
 1 | from __future__ import print_function
 2 | 
 3 | import cv2
 4 | import numpy as np
 5 | import tensorflow as tf
 6 | import socket
 7 | 
 8 | video_output = True
 9 | 
10 | from time import time
11 | time_next  = time()
12 | 
13 | if video_output:
14 |   time_delay = 0.2
15 | else:
16 |   time_delay = 0.05
17 | 
18 | import train
19 | 
20 | udp_host = "localhost"
21 | udp_port = 4000
22 | sock     = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
23 | 
24 | key_wait_time = 10
25 | input_size = 128
26 | 
27 | capture = cv2.VideoCapture(0)
28 | 
29 | # video_size = (int(capture.get(cv2.CAP_PROP_FRAME_WIDTH)),
30 | #               int(capture.get(cv2.CAP_PROP_FRAME_HEIGHT)))
31 | # print "Video width, height: " + str(video_size)
32 | 
33 | model = train.build_model()
34 | model.load('checkpoints/road_model1-360')
35 | 
36 | def process_frame(frame):
37 |   pr = model.predict(frame[np.newaxis, :, :, np.newaxis])
38 |   return pr[0][0]
39 | 
40 | while capture.isOpened():
41 |   success, frame = capture.read()
42 | 
43 |   if success:
44 |     time_now = time()
45 |     if time_now >= time_next:
46 |       time_next = time_now + time_delay
47 | 
48 |       frame_gray  = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
49 |       frame_small = cv2.resize(frame_gray, (128, 128))
50 | #     frame_small = cv2.resize(frame_gray, (128, 128 + 95))
51 |       frame_small = cv2.flip(frame_small, 0)  # horizontal flip
52 |       frame_small = cv2.flip(frame_small, 1)  # vertical flip
53 | #     frame_small = frame_small[-128:,]
54 |       if video_output:
55 |         cv2.imshow('video', frame_small)
56 | 
57 |       output_value =  process_frame(frame_small)
58 |       print(output_value)
59 |       sock.sendto('ai: %.6f' % output_value, (udp_host, udp_port))
60 | 
61 |     ch = cv2.waitKey(key_wait_time) & 0xFF
62 |     if ch == 27:
63 |       break
64 |     if ch == ord('q'):
65 |       break
66 | 
67 | capture.release()
68 | cv2.destroyAllWindows()
69 | 


--------------------------------------------------------------------------------
/src/nodebot/nodebot.js:
--------------------------------------------------------------------------------
  1 | #!/usr/bin/env node
  2 | 
  3 | /*
  4 |  * Usage
  5 |  * ~~~~~
  6 |  *   node nodebot.js            # keyboard enabled,  search for port
  7 |  *   node nodebot.js stdin_off  # keyboard disabled, search for port
  8 |  *   node nodebot.js port       # specific Johnny-Five board port
  9 |  *
 10 |  * Operating
 11 |  * ~~~~~~~~~
 12 |  * - LED matrix: pin 14 clock, pin 15 data
 13 |  * - Ultrasonic sensor (ping): pin A3
 14 |  *
 15 |  * To Do
 16 |  * ~~~~~
 17 |  * - Implement "h" for help
 18 |  * - LED screen: Motor speed as vertical bars
 19 |  * - LED screen: AI output as horizontal bar
 20 |  * - LED screen: Current state: f, b, l, r, s, ai on / off
 21 |  *
 22 |  * - Implement line following: reflectance sensors
 23 |  * - Implement LEDs (pin 7 ?)
 24 |  * - Implement button sensor
 25 |  * - Implement light sensor
 26 |  * - Implement buzzer
 27 |  *
 28 |  * Resources
 29 |  * ~~~~~~~~~
 30 |  * github.com/Makeblock-official/mbot_nodebots/blob/master/examples/button.js
 31 |  *
 32 |  * github.com/Makeblock-official/mbot_nodebots/blob/master/examples/leds.js
 33 |  *
 34 |  * github.com/Makeblock-official/mbot_nodebots/blob/master/examples/light.js
 35 |  *
 36 |  * github.com/rwaldron/johnny-five/wiki/Motor
 37 |  * github.com/Makeblock-official/mbot_nodebots/blob/master/examples/motors.js
 38 |  *
 39 |  * github.com/Makeblock-official/mbot_nodebots/blob/master/examples/piezo.js
 40 |  *
 41 |  * github.com/Makeblock-official/mbot_nodebots/blob/master/examples/reflectance.js
 42 |  *
 43 |  * github.com/rwaldron/johnny-five/wiki/Proximity
 44 |  * github.com/Makeblock-official/mbot_nodebots/blob/master/examples/sonar.js
 45 |  * https://gist.githubusercontent.com/rwaldron/0519fcd5c48bfe43b827/raw/f17fb09b92ed04722953823d9416649ff380c35b/PingFirmata.ino
 46 |  */
 47 | 
 48 | var board_port;
 49 | var stdin_off = false;
 50 | 
 51 | if (process.argv.length === 3) {
 52 |   if (process.argv[2] === 'stdin_off') {
 53 |     stdin_off = true;
 54 |   }
 55 |   else {
 56 |     board_port = process.argv[2]
 57 |   }
 58 | }
 59 | 
 60 | var UDP_PORT = 4000;
 61 | 
 62 | var ai_fudge_factor = 0.011;
 63 | 
 64 | var proximity_limit = 20.0;  // centimeters
 65 | var proximity_last  =  0.0;
 66 | 
 67 | var speed_min    =  80;
 68 | var speed_max    = 120;
 69 | var speed_factor = 0.5;
 70 | var speed_motor  = speed_factor * (speed_max - speed_min) + speed_min;
 71 | 
 72 | var dgram       = require('dgram');
 73 | var johnny_five = require('johnny-five');
 74 | var led_screen  = require('./tm1640_led_screen');
 75 | 
 76 | console.log('NodeBot: Connecting');
 77 | var board = new johnny_five.Board({port: board_port});
 78 | 
 79 | var motor_left, motor_right;
 80 | var screen;
 81 | var state    = 'setup';
 82 | var state_ai = false;
 83 | 
 84 | board.on('ready', function(error) {
 85 |   if (error) {
 86 |     console.log(error);
 87 |     return;
 88 |   }
 89 | 
 90 |   console.log('NodeBot: Connected');
 91 | 
 92 |   screen = led_screen.initialize(johnny_five, board, 14, 15);
 93 |   led_screen.clear_screen(screen);
 94 |   led_screen.draw_character(screen, 6, 1, '?');
 95 |   led_screen.write_screen(screen);
 96 | 
 97 |   motor_left = new johnny_five.Motor({
 98 |     pins: { pwm: 6, dir: 7 }
 99 |   });
100 | 
101 |   motor_right = new johnny_five.Motor({
102 |     pins: { pwm: 5, dir: 4 }
103 |   });
104 | 
105 |   motor_left.stop();
106 |   motor_right.stop();
107 |   state = 'stop';
108 |   console.log('State: ' + state + ', ai: ' + state_ai);
109 | 
110 |   var proximity = new johnny_five.Proximity({
111 |     freq:       250,
112 |     controller: 'HCSR04',
113 |     pin:        'A3'
114 |   });
115 | 
116 |   proximity.on('data', function() {
117 |     var proximity = this.cm;
118 | 
119 |     var proximity_delta = Math.abs(proximity - proximity_last);
120 |     if (proximity <= 100  &&  proximity_delta >=  1.0  ||
121 |         proximity >= 100  &&  proximity_delta >= 10.0) {
122 | 
123 |       console.log('cm: ' + proximity.toFixed(0));
124 |       proximity_last = proximity;
125 |     }
126 | 
127 |     if (state === 'run'  &&  proximity < proximity_limit) {
128 |       motor_left.stop();
129 |       motor_right.stop();
130 |       state = 'pause';
131 |       console.log('State: ' + state + ', ai: ' + state_ai);
132 |     }
133 | 
134 |     if (state === 'pause'  &&  proximity > proximity_limit) {
135 |       motor_left.reverse(speed_motor);
136 |       motor_right.forward(speed_motor);
137 |       state = 'run';
138 |       console.log('State: ' + state + ', ai: ' + state_ai);
139 |     }
140 |   });
141 | 
142 |   console.log('NodeBot: Ready');
143 | });
144 | 
145 | function action(command) {
146 |   if (state === 'setup') return;
147 | 
148 |   if (command >= '0'  &&  command <= '9') {
149 |     value = parseInt(command);
150 |     if (value === 0) value = 10;
151 |     speed_factor = value / 10;
152 |     speed_motor  = speed_factor * (speed_max - speed_min) + speed_min;
153 | 
154 |     motor_left.reverse(speed_motor);
155 |     motor_right.forward(speed_motor);
156 |     state = 'run';
157 |   }
158 | 
159 |   switch (command) {
160 |     case ' ':
161 |       motor_left.stop();
162 |       motor_right.stop();
163 |       state = 'stop';
164 |       console.log('State: ' + state + ', ai: ' + state_ai);
165 |       break;
166 | 
167 |     case 'a':
168 |       state_ai = ! state_ai;
169 |       console.log('State: ' + state + ', ai: ' + state_ai);
170 |       break;
171 | 
172 |     case 'f':
173 |       motor_left.reverse(speed_motor);
174 |       motor_right.forward(speed_motor);
175 |       state = 'run';
176 |       console.log('State: ' + state + ', ai: ' + state_ai);
177 |       break;
178 | 
179 |     case 'b':
180 |       motor_left.forward(speed_motor);
181 |       motor_right.reverse(speed_motor);
182 |       state = 'run';
183 |       console.log('Turn: back');
184 |       break;
185 | 
186 |     case 'l':
187 |       motor_left.forward(speed_motor);
188 |       motor_right.forward(speed_motor);
189 |       state = 'run';
190 |       console.log('Turn: left');
191 |       break;
192 | 
193 |     case 'r':
194 |       motor_left.reverse(speed_motor);
195 |       motor_right.reverse(speed_motor);
196 |       state = 'run';
197 |       console.log('Turn: right');
198 |       break;
199 | 
200 |     case 'q':
201 |       console.log('Exiting');
202 |       motor_left.stop();
203 |       motor_right.stop();
204 |       state = 'stop';
205 |       process.exit();
206 |       break;
207 | 
208 |     case 'mood:clear':
209 |       led_screen.clear_screen(screen);
210 |       led_screen.write_screen(screen);
211 |       break;
212 | 
213 |     case 'mood:happy':
214 |       screen.matrix = new Buffer([
215 |         0x00, 0x00, 0x00, 0x00, 0x00, 0x48, 0x88, 0x80,
216 |         0x80, 0x88, 0x48, 0x00, 0x00, 0x00, 0x00, 0x00
217 |       ]);
218 |       led_screen.write_screen(screen);
219 |       break;
220 | 
221 |     case 'mood:neutral':
222 |       screen.matrix = new Buffer([
223 |         0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x48, 0x40,
224 |         0x40, 0x48, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
225 |       ]);
226 |       led_screen.write_screen(screen);
227 |       break;
228 | 
229 |     case 'mood:sad':
230 |       screen.matrix = new Buffer([
231 |         0x00, 0x00, 0x00, 0x00, 0x00, 0x80, 0x48, 0x40,
232 |         0x40, 0x48, 0x80, 0x00, 0x00, 0x00, 0x00, 0x00
233 |       ]);
234 |       led_screen.write_screen(screen);
235 |       break;
236 |   }
237 | }
238 | 
239 | /*
240 |  * There is no stdin when running in the background,
241 |  * then create a stub stdin that does nothing.
242 |  */
243 | 
244 | if (stdin_off) {
245 |   var Readable = require("stream").Readable;
246 |   var util     = require("util");
247 | 
248 |   function MyStream(options) {
249 |     Readable.call(this, options);
250 |   }
251 |   MyStream.prototype._read = function() {};
252 | 
253 |   util.inherits(MyStream, Readable);
254 | 
255 |   process.__defineGetter__("stdin", function() {
256 |     if (process.__stdin) return(process.__stdin);
257 |     process.__stdin = new MyStream();
258 |     return(process.__stdin);
259 |   });
260 | }
261 | else {
262 |   var stdin = process.stdin;
263 |   stdin.setRawMode(true);
264 |   stdin.resume();
265 |   stdin.setEncoding('utf8');
266 | 
267 |   stdin.on('data', function(key) {
268 |     action(key);
269 |   });
270 | }
271 | 
272 | var server = dgram.createSocket('udp4');
273 | var ai_output_last = 0.0;
274 | 
275 | server.on('message', function (message, rinfo) {
276 | //console.log('Received: ' + rinfo.address + ':' + rinfo.port + ': ' + message);
277 | //console.log('Received: ' + message);
278 | 
279 |   message = message.toString();
280 | 
281 |   if (message.startsWith('ai:')) {
282 |     if (state === 'run'  &&  state_ai === true) {
283 |       var ai_output = parseFloat(message.substring(3)) - ai_fudge_factor;
284 | 
285 |       if (Math.abs(ai_output - ai_output_last) >= 0.0) {
286 |         ai_output_last = ai_output;
287 | 
288 |         var ai_motor_left  = (-1.666 *  ai_output) + 0.5;
289 |         var ai_motor_right =   1.666 * (ai_output + 0.3);
290 | 
291 |         var precision = 2;
292 |         console.log('ai: ' + ai_output.toFixed(precision)      +
293 |                   ', ml: ' + ai_motor_left.toFixed(precision)  +
294 |                   ', mr: ' + ai_motor_right.toFixed(precision) +
295 |                   ', ms: ' + speed_motor);
296 | 
297 |         motor_left.reverse(ai_motor_left   * speed_motor * 2);
298 |         motor_right.forward(ai_motor_right * speed_motor * 2);
299 |       }
300 |     }
301 |   }
302 |   else if (message.startsWith('mood:')) {
303 |     action(message.substring(0, message.length -1));  // remove newline
304 |   }
305 |   else {
306 |     action(message.substring(0,1));  // use first character
307 |   }
308 | });
309 | 
310 | server.bind(UDP_PORT);
311 | 


--------------------------------------------------------------------------------
/src/nodebot/package.json:
--------------------------------------------------------------------------------
 1 | {
 2 |   "name": "nodebot-ai",
 3 |   "version": "0.1.0",
 4 |   "description": "Nodebots code to support ai elements",
 5 |   "main": "nodebot.js",
 6 |   "scripts": {
 7 |     "test": "echo \"Error: no test specified\" && exit 1"
 8 |   },
 9 |   "author": "",
10 |   "license": "ISC",
11 |   "dependencies": {
12 |     "johnny-five": "^0.10.0",
13 |     "nodebots-interchange": "^1.1.2"
14 |   }
15 | }
16 | 


--------------------------------------------------------------------------------
/src/nodebot/tm1640_led_screen.js:
--------------------------------------------------------------------------------
  1 | #!/usr/bin/env node
  2 | 
  3 | /*
  4 |  * Description
  5 |  * ~~~~~~~~~~~
  6 |  * NodeBots (Johnny-Five) support for MakeBlock LED Matrix peripheral.
  7 |  * Contains a simple driver for TM1640 16 x 8 LED driver IC.
  8 |  * Provides simple low-level graphics support, e.g clear screen,
  9 |  *   draw point, line,horizontal line, vertical line, rectangle and cirle.
 10 |  *
 11 |  * Author(s): Andrew Fisher (@ajfisher)  https://github.com/ajfisher
 12 |  *            Andy Gelme    (@geekscape) https://github.com/geekscape
 13 |  *
 14 |  * Requirements
 15 |  * ~~~~~~~~~~~~
 16 |  * An Arduino running Firmata with a MakeBlock TM1640 LED Matrix.
 17 |  *
 18 |  *   npm install johnny-five
 19 |  *   npm install oled-font-5x7
 20 |  *
 21 |  * Usage: NodeBot (NodeJS + Johnny-Five) module
 22 |  * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 23 |  * Application code must maintain a reference to the "screen" variable.
 24 |  *
 25 |  *   var five       = require('johnny-five');
 26 |  *   var board      = five.Board();
 27 |  *   var led_screen = require('./tm1640_led_screen');
 28 |  *
 29 |  *   board.on('ready', function() {
 30 |  *     var screen = led_screen.initialize(five, board, 14, 15);
 31 |  *     led_screen.clear_screen(screen);
 32 |  *     led_screen.draw_character(screen, 6, 0, '?');
 33 |  *     led_screen.write_screen(screen);
 34 |  *     screen.matrix = new Buffer([     // Checkerboard
 35 |  *       0xaa, 0x55, 0xaa, 0x55, 0xaa, 0x55, 0xaa, 0x55,
 36 |  *       0xaa, 0x55, 0xaa, 0x55, 0xaa, 0x55, 0xaa, 0x55
 37 |  *     ]);
 38 |  *     led_screen.write_screen(screen);
 39 |  *   });
 40 |  *
 41 |  * Usage: Command line testing using NodeBots (Johnny-Five) REPL
 42 |  * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 43 |  * This test environment automatically maintains an internal "screen" variable.
 44 |  *
 45 |  *   $ node tm1640_led_screen.js test
 46 |  *   1469988451381 Device(s) /dev/cu.wchusbserial1410
 47 |  *   1469988451391 Connected /dev/cu.wchusbserial1410
 48 |  *   1469988455308 Repl Initialized
 49 |  *   NodeBot ready, type "help()" for help !
 50 |  *   >> help();
 51 |  *   >> test();  // stop test running
 52 |  *   >> clear_screen();
 53 |  *   >> draw_character(6, 0, '?');
 54 |  *
 55 |  * Resources
 56 |  * ~~~~~~~~~
 57 |  * Titan Micro Electronics TM1640: 16 x 8 LED driver datasheet (Chinese)
 58 |  * https://dl.dropboxusercontent.com/u/8663580/TM1640.pdf
 59 |  *
 60 |  * To Do
 61 |  * ~~~~~
 62 |  * - Check all function parameters are within range (bounds).
 63 |  */
 64 | 
 65 | var WIDTH  = 16;
 66 | var HEIGHT =  8;
 67 | 
 68 | var font = require('oled-font-5x7');
 69 | 
 70 | var test_enabled = (process.argv.length === 3  &&  process.argv[2] === 'test');
 71 | 
 72 | if (test_enabled) {
 73 |   var five  = require('johnny-five');
 74 |   var board = five.Board();
 75 | 
 76 |   board.on('ready', function() {
 77 |     console.log('NodeBot ready, type "help()" for help !');
 78 | 
 79 |     var PIN_CLOCK = 14;
 80 |     var PIN_DATA  = 15;
 81 |     var screen    = initialize(five, board, PIN_CLOCK, PIN_DATA);
 82 | 
 83 |     clear_screen(screen);
 84 | 
 85 |     if (true) {
 86 |       draw_character(screen,  1, 0, 'G');
 87 |       draw_character(screen,  7, 0, 'o');
 88 |       draw_character(screen, 13, 0, '!');
 89 |       write_screen(screen);
 90 |     }
 91 |     else {
 92 |       screen.matrix = new Buffer([     // Checkerboard
 93 |         0xaa, 0x55, 0xaa, 0x55, 0xaa, 0x55, 0xaa, 0x55,
 94 |         0xaa, 0x55, 0xaa, 0x55, 0xaa, 0x55, 0xaa, 0x55
 95 |       ]);
 96 |       write_screen(screen);
 97 |     }
 98 | 
 99 |     clear_screen(screen);
100 | 
101 |     var state_running = true;
102 |     var state_visible = false;
103 | 
104 |     setInterval(function() {
105 |       if (state_running) {
106 |         state_visible = ! state_visible;
107 | 
108 |         if (true) {                                         // Face ?
109 |           draw_rectangle(screen, 0, 0, WIDTH, HEIGHT, state_visible);
110 |           draw_circle(screen,  5, 3, 1, state_visible);
111 |           draw_circle(screen, 10, 3, 1, state_visible);
112 |           draw_lineh(screen,   7, 5, 2, state_visible);
113 |         }
114 |         else {                                     // X marks the spot !
115 |           draw_line(screen, 0, 0, H - 1, HEIGHT - 1,     state_visible);
116 |           draw_line(screen, 0, HEIGHT - 1, WIDTH - 1, 0, state_visible);
117 |         }
118 | 
119 | //      invert_screen(screen);
120 | 
121 |         write_screen(screen);
122 |       }
123 |     }.bind(this), 1000);
124 | 
125 |     this.repl.inject({
126 |       help: function() {
127 |         console.log('Columns: 0 - 15, rows, 0 - 7, optional value: 0 - 1');
128 |         console.log('Functions:');
129 |         console.log('  clear_screen(value)');
130 |         console.log('  draw_character(column, row, character, value)');
131 |         console.log('  draw_circle(column, row, radius, value)');
132 |         console.log('  draw_line(column0, row0, column1, row1, value)');
133 |         console.log('  draw_lineh(column, row, length, value)');
134 |         console.log('  draw_linev(column, row, length, value)');
135 |         console.log('  draw_point(column, row, value)');
136 |         console.log('  draw_rectangle(column, row, width, height, value)');
137 |         console.log('  fill_circle(column, row, radius, value)');
138 |         console.log('  fill_rectangle(column, row, width, height, value)');
139 |         console.log('  test(): Toggle between test running or not');
140 |       },
141 |       clear_screen: function(value) {
142 |         clear_screen(screen, value);
143 |       },
144 |       draw_character: function(column, row, character, value) {
145 |         draw_character(screen, column, row, character, value);
146 |         write_screen(screen);
147 |       },
148 |       draw_circle: function(column, row, radius, value) {
149 |         draw_circle(screen, column, row, radius, value);
150 |         write_screen(screen);
151 |       },
152 |       draw_line: function(column0, row0, column1, row1, value) {
153 |         draw_line(screen, column0, row0, column1, row1, value);
154 |         write_screen(screen);
155 |       },
156 |       draw_lineh: function(column, row, length, value) {
157 |         draw_lineh(screen, column, row, length, value);
158 |         write_screen(screen);
159 |       },
160 |       draw_linev: function(column, row, length, value) {
161 |         draw_linev(screen, column, row, length, value);
162 |         write_screen(screen);
163 |       },
164 |       draw_point: function(column, row, value) {
165 |         draw_point(screen, column, row, value);
166 |         write_screen(screen);
167 |       },
168 |       draw_rectangle: function(column, row, width, height, value) {
169 |         draw_rectangle(screen, column, row, width, height, value);
170 |         write_screen(screen);
171 |       },
172 |       fill_circle: function(column, row, radius, value) {
173 |         fill_circle(screen, column, row, radius, value);
174 |         write_screen(screen);
175 |       },
176 |       fill_rectangle: function(column, row, width, height, value) {
177 |         draw_rectangle(screen, column, row, width, height, value);
178 |         write_screen(screen);
179 |       },
180 |       invert_screen: function(value) {
181 |         invert_screen(screen);
182 |         write_screen(screen);
183 |       },
184 |       test: function(value) {
185 |         state_running = ! state_running;
186 |       console.log('Test run state: ' + state_running);
187 |       }
188 |     });
189 |   });
190 | }
191 | 
192 | function initialize(five, board, pin_clock_number, pin_data_number) {
193 |   var MODE_ADDRESS_AUTO_ADD_1 = 0x40;
194 |   var MODE_PERMANENT_ADDRESS  = 0x44;
195 | 
196 |   var pin_clock = five.Pin({
197 |     pin:  pin_clock_number,
198 |     mode: board.io.MODES.OUTPUT,
199 |   });
200 | 
201 |   var pin_data = five.Pin({
202 |     pin:  pin_data_number,
203 |     mode: board.io.MODES.OUTPUT,
204 |   });
205 | 
206 |   write_byte(pin_clock, pin_data, MODE_ADDRESS_AUTO_ADD_1);
207 |   write_byte(pin_clock, pin_data, 0x8c);
208 | 
209 |   var screen = {
210 |     pin_clock: pin_clock, pin_data: pin_data, matrix: new Buffer(16)
211 |   };
212 | 
213 |   return(screen);
214 | }
215 | 
216 | function write_byte(pin_clock, pin_data, buffer) {
217 |   pin_clock.high();  pin_data.low();
218 | 
219 |   for (var bit = 0;  bit < 8; bit ++) {
220 |     pin_clock.low();
221 |     pin_data.write(buffer & 0x01);
222 |     pin_clock.high();
223 |     buffer = buffer >> 1;
224 |   }
225 | 
226 |   pin_clock.low();   pin_data.low();
227 |   pin_clock.high();  pin_data.high();
228 | }
229 | 
230 | function write_bytes_to_address(pin_clock, pin_data, address, buffer) {
231 |   address = address | 0xc0;
232 | 
233 |   pin_clock.high();  pin_data.low();
234 | 
235 |   for (var bit = 0;  bit < 8; bit ++) {
236 |     pin_clock.low();
237 |     pin_data.write(address & 0x01);
238 |     pin_clock.high();
239 |     address = address >> 1;
240 |   }
241 | 
242 |   for (var byte = 0;  byte < buffer.length;  byte ++) {
243 |     var data = buffer[byte];
244 | 
245 |     for (bit = 0; bit < 8; bit ++) {
246 |       pin_clock.low();
247 |       pin_data.write(data & 0x01);
248 |       pin_clock.high();
249 |       data = data >> 1;
250 |     }
251 |   }
252 | 
253 |   pin_clock.low();   pin_data.low();
254 |   pin_clock.high();  pin_data.high();
255 | }
256 | 
257 | function write_screen(screen) {
258 |   write_bytes_to_address(screen.pin_clock, screen.pin_data, 0, screen.matrix);
259 | }
260 | 
261 | function clear_screen(screen, value) {  // value = 0xff for all LEDS on
262 |   if (typeof(value) === 'undefined') value = 0;
263 |   screen.matrix.fill(value);
264 |   write_screen(screen);
265 | }
266 | 
267 | function invert_screen(screen) {
268 |   for (var index = 0;  index < screen.matrix.length;  index ++) {
269 |     screen.matrix[index] = screen.matrix[index] ^ 0xff;
270 |   }
271 | }
272 | 
273 | function draw_point(screen, column, row, value) {
274 |   var bit  = (typeof(value) === 'undefined')  ?  1  :  value;
275 |   var mask = 0xff ^ (1 << row);
276 |   screen.matrix[column] = screen.matrix[column] & mask | (bit << row);
277 | 
278 | // TODO: This should work and would be much more efficient !
279 | //write_bytes_to_address(
280 | //  screen.pin_clock, screen.pin_data, column, new Buffer(screen.matrix[column])
281 | //);
282 | }
283 | 
284 | // Bresenham's line algorithm
285 | 
286 | function draw_line(screen, column0, row0, column1, row1, value) {
287 |   var column_delta = Math.abs(column1 - column0);
288 |   var row_delta    = Math.abs(row1    - row0);
289 | 
290 |   var column_increment = column0 < column1 ? 1 : -1;
291 |   var row_increment    = row0    < row1    ? 1 : -1;
292 | 
293 |   var error = (column_delta > row_delta ? column_delta : -row_delta) / 2;
294 | 
295 |   while (true) {
296 |     draw_point(screen, column0, row0, value);
297 | 
298 |     if (column0 === column1  &&  row0 === row1) break;
299 | 
300 |     var error2 = error;
301 | 
302 |     if (error2 > -column_delta) {
303 |       error   -= row_delta;
304 |       column0 += column_increment;
305 |     }
306 | 
307 |     if (error2 < row_delta) {
308 |       error += column_delta;
309 |       row0  += row_increment;
310 |     }
311 |   }
312 | }
313 | 
314 | function draw_lineh(screen, column, row, length, value) {
315 |   for (var index = column;  index < column + length;  index ++) {
316 |     draw_point(screen, index, row, value);
317 |   }
318 | 
319 | // TODO: More efficient to write_bytes_to_address() for only changed columns !
320 | }
321 | 
322 | function draw_linev(screen, column, row, length, value) {
323 |   var bit  = (typeof(value) === 'undefined')  ?  1  :  value;
324 |   var byte = (Math.pow(2, length) - 1) << row;
325 |   var mask = 0xff ^ byte;
326 |   if (bit === false) byte = 0;
327 |   screen.matrix[column] = screen.matrix[column] & mask | byte;
328 | 
329 | // TODO: This should work and would be much more efficient !
330 | //write_bytes_to_address(
331 | //  screen.pin_clock, screen.pin_data, column, new Buffer(screen.matrix[column])
332 | //);
333 | }
334 | 
335 | function draw_rectangle(screen, row, column, width, height, value) {
336 |   draw_lineh(screen, column, row,              width,  value);
337 |   draw_lineh(screen, column, row + height - 1, width,  value);
338 |   draw_linev(screen, column, row,              height, value);
339 |   draw_linev(screen, column + width - 1, row,  height, value);
340 | }
341 | 
342 | function fill_rectangle(screen, column, row, width, height, value) {
343 |   for (var index = 0;  index < width;  index ++) {
344 |     draw_linev(screen, column + index, row, height, value);
345 |   }
346 | }
347 | 
348 | // Bresenham's line algorithm extended for circles
349 | 
350 | function draw_circle(screen, column, row, radius, value) {
351 |   var x = radius, y = 0;
352 |   var radiusError = 1 - x;
353 | 
354 |   while (x >= y) {
355 |     draw_point(screen, -y + column, -x + row, value);
356 |     draw_point(screen,  y + column, -x + row, value);
357 |     draw_point(screen, -x + column, -y + row, value);
358 |     draw_point(screen,  x + column, -y + row, value);
359 |     draw_point(screen, -x + column,  y + row, value);
360 |     draw_point(screen,  x + column,  y + row, value);
361 |     draw_point(screen, -y + column,  x + row, value);
362 |     draw_point(screen,  y + column,  x + row, value);
363 |     y++;
364 | 
365 |     if (radiusError < 0) {
366 |       radiusError += 2 * y + 1;
367 |     }
368 |     else {
369 |       x --;
370 |       radiusError += 2 * (y - x + 1);
371 |     }
372 |   }
373 | }
374 | 
375 | function fill_circle(screen, column, row, radius, value) {
376 |   var x = radius, y = 0;
377 |   var radiusError = 1 - x;
378 | 
379 |   while (x >= y) {
380 |     draw_line(screen, -y + column, -x + row, y + column, -x + row, value);
381 |     draw_line(screen, -x + column, -y + row, x + column, -y + row, value);
382 |     draw_line(screen, -x + column,  y + row, x + column,  y + row, value);
383 |     draw_line(screen, -y + column,  x + row, y + column,  x + row, value);
384 |     y++;
385 | 
386 |     if (radiusError < 0) {
387 |       radiusError += 2 * y + 1;
388 |     }
389 |     else {
390 |       x --;
391 |       radiusError+= 2 * (y - x + 1);
392 |     }
393 |   }
394 | }
395 | 
396 | // Thanks Suz (@noopkat) !
397 | // https://github.com/noopkat/oled-font-5x7
398 | 
399 | function draw_character(screen, column, row, character, value) {
400 |   var lookup   = index = font.lookup.indexOf(character) * 5;
401 |   var fontData = font.fontData.slice(lookup, lookup + 5);
402 | 
403 |   for (var index = 0;  index < fontData.length;  index ++) {
404 |     screen.matrix[column + index] = fontData[index] << 1;
405 |   }
406 | }
407 | 
408 | module.exports = {
409 |   initialize:             initialize,
410 |   write_byte:             write_byte,
411 |   write_bytes_to_address: write_bytes_to_address,
412 |   write_screen:           write_screen,
413 |   clear_screen:           clear_screen,
414 |   invert_screen:          invert_screen,
415 |   draw_point:             draw_point,
416 |   draw_line:              draw_line,
417 |   draw_lineh:             draw_lineh,
418 |   draw_linev:             draw_linev,
419 |   draw_rectangle:         draw_rectangle,
420 |   fill_rectangle:         fill_rectangle,
421 |   draw_circle:            draw_circle,
422 |   fill_circle:            fill_circle,
423 |   draw_character:         draw_character
424 | };
425 | 


--------------------------------------------------------------------------------
/src/nodebot/tm1640_test.js:
--------------------------------------------------------------------------------
 1 | #!/usr/bin/env node
 2 | 
 3 | var five       = require('johnny-five');
 4 | var board      = five.Board();
 5 | var led_screen = require('./tm1640_led_screen');
 6 | 
 7 | board.on('ready', function() {
 8 |   console.log('NodeBot ready');
 9 | 
10 |   var screen = led_screen.initialize(five, board, 14, 15);
11 | 
12 |   led_screen.clear_screen(screen);
13 |   led_screen.draw_character(screen, 6, 0, '?');
14 |   led_screen.write_screen(screen);
15 | 
16 | //screen.matrix = new Buffer([     // Checkerboard
17 | //  0xaa, 0x55, 0xaa, 0x55, 0xaa, 0x55, 0xaa, 0x55,
18 | //  0xaa, 0x55, 0xaa, 0x55, 0xaa, 0x55, 0xaa, 0x55
19 | //]);
20 | //led_screen.write_screen(screen);
21 | });
22 | 


--------------------------------------------------------------------------------
/src/nodebot/udp_receive.js:
--------------------------------------------------------------------------------
 1 | #!/usr/bin/env node
 2 | 
 3 | var dgram = require('dgram');
 4 | 
 5 | var UDP_PORT = 4000;
 6 | 
 7 | var server = dgram.createSocket('udp4');
 8 | 
 9 | server.on('message', function (msg, rinfo) {
10 |   console.log('Received: ' + rinfo.address + ':' + rinfo.port + ': ' + msg);
11 | });
12 | 
13 | server.bind(UDP_PORT);
14 | 


--------------------------------------------------------------------------------