Mercurial > cortex
comparison thesis/dylan-cortex-diff.diff @ 514:447c3c8405a2
accept/reject changes
author | Robert McIntyre <rlm@mit.edu> |
---|---|
date | Sun, 30 Mar 2014 10:50:05 -0400 |
parents | 4c4d45f6f30b |
children | 58fa1ffd481e |
comparison
equal
deleted
inserted
replaced
513:4c4d45f6f30b | 514:447c3c8405a2 |
---|---|
292 -** Because of Time, simulation is perferable to reality | 292 -** Because of Time, simulation is perferable to reality |
293 +** Simulated time enables rapid prototyping and complex scenes | 293 +** Simulated time enables rapid prototyping and complex scenes |
294 | 294 |
295 I envision =CORTEX= being used to support rapid prototyping and | 295 I envision =CORTEX= being used to support rapid prototyping and |
296 iteration of ideas. Even if I could put together a well constructed | 296 iteration of ideas. Even if I could put together a well constructed |
297 @@ -459,8 +577,8 @@ | |
298 simulations of very simple creatures in =CORTEX= generally run at | |
299 40x on my machine! | |
300 | |
301 -** What is a sense? | |
302 - | |
303 +** All sense organs are two-dimensional surfaces | |
304 +# What is a sense? | |
305 If =CORTEX= is to support a wide variety of senses, it would help | |
306 to have a better understanding of what a ``sense'' actually is! | |
307 While vision, touch, and hearing all seem like they are quite | |
308 @@ -956,7 +1074,7 @@ | |
309 #+ATTR_LaTeX: :width 15cm | |
310 [[./images/physical-hand.png]] | |
311 | |
312 -** Eyes reuse standard video game components | |
313 +** Sight reuses standard video game components... | |
314 | |
315 Vision is one of the most important senses for humans, so I need to | |
316 build a simulated sense of vision for my AI. I will do this with | |
317 @@ -1257,8 +1375,8 @@ | |
318 community and is now (in modified form) part of a system for | |
319 capturing in-game video to a file. | |
320 | |
321 -** Hearing is hard; =CORTEX= does it right | |
322 - | |
323 +** ...but hearing must be built from scratch | |
324 +# is hard; =CORTEX= does it right | |
325 At the end of this section I will have simulated ears that work the | |
326 same way as the simulated eyes in the last section. I will be able to | |
327 place any number of ear-nodes in a blender file, and they will bind to | |
328 @@ -1565,7 +1683,7 @@ | |
329 jMonkeyEngine3 community and is used to record audio for demo | |
330 videos. | |
331 | |
332 -** Touch uses hundreds of hair-like elements | |
333 +** Hundreds of hair-like elements provide a sense of touch | |
334 | |
335 Touch is critical to navigation and spatial reasoning and as such I | |
336 need a simulated version of it to give to my AI creatures. | |
337 @@ -2059,7 +2177,7 @@ | |
338 #+ATTR_LaTeX: :width 15cm | |
339 [[./images/touch-cube.png]] | |
340 | |
341 -** Proprioception is the sense that makes everything ``real'' | |
342 +** Proprioception provides knowledge of your own body's position | |
343 | |
344 Close your eyes, and touch your nose with your right index finger. | |
345 How did you do it? You could not see your hand, and neither your | |
346 @@ -2193,7 +2311,7 @@ | |
347 #+ATTR_LaTeX: :width 11cm | |
348 [[./images/proprio.png]] | |
349 | |
350 -** Muscles are both effectors and sensors | |
351 +** Muscles contain both sensors and effectors | |
352 | |
353 Surprisingly enough, terrestrial creatures only move by using | |
354 torque applied about their joints. There's not a single straight | |
355 @@ -2440,7 +2558,8 @@ | 297 @@ -2440,7 +2558,8 @@ |
356 hard control problems without worrying about physics or | 298 hard control problems without worrying about physics or |
357 senses. | 299 senses. |
358 | 300 |
359 -* Empathy in a simulated worm | 301 -* Empathy in a simulated worm |
369 -** Empathy is the process of tracing though \Phi-space | 311 -** Empathy is the process of tracing though \Phi-space |
370 +** ``Empathy'' requires retracing steps though \Phi-space | 312 +** ``Empathy'' requires retracing steps though \Phi-space |
371 | 313 |
372 Here is the core of a basic empathy algorithm, starting with an | 314 Here is the core of a basic empathy algorithm, starting with an |
373 experience vector: | 315 experience vector: |
374 @@ -2888,7 +3007,7 @@ | |
375 #+end_src | |
376 #+end_listing | |
377 | |
378 -** Efficient action recognition with =EMPATH= | |
379 +** =EMPATH= recognizes actions efficiently | |
380 | |
381 To use =EMPATH= with the worm, I first need to gather a set of | |
382 experiences from the worm that includes the actions I want to | |
383 @@ -3044,9 +3163,9 @@ | |
384 to interpretation, and dissaggrement between empathy and experience | |
385 is more excusable. | |
386 | |
387 -** Digression: bootstrapping touch using free exploration | |
388 - | |
389 - In the previous section I showed how to compute actions in terms of | |
390 +** Digression: Learn touch sensor layout through haptic experimentation, instead | |
391 +# Boostraping touch using free exploration | |
392 +In the previous section I showed how to compute actions in terms of | |
393 body-centered predicates which relied averate touch activation of | |
394 pre-defined regions of the worm's skin. What if, instead of recieving | |
395 touch pre-grouped into the six faces of each worm segment, the true |