annotate thesis/abstract.org @ 435:ae3bfc82ac7c

on dylan's suggestion, rewriting abstract to show hand more.
author Robert McIntyre <rlm@mit.edu>
date Sun, 23 Mar 2014 17:59:16 -0400
parents d52bff980f0d
children 853377051f1e
rev   line source
rlm@432 1 Here I explore the design and capabilities of my system (called
rlm@432 2 =CORTEX=) which enables experiments in /embodied artificial
rlm@432 3 intelligence/ -- that is, AI which uses a physical simulation of
rlm@432 4 reality accompanied by a simulated body to solve problems.
rlm@432 5
rlm@432 6 In the first half of the thesis I describe the construction of
rlm@432 7 =CORTEX= and the rationale behind my architecture choices. =CORTEX= is
rlm@432 8 a complete platform for embodied AI research. It provides multiple
rlm@432 9 senses for simulated creatures, including vision, touch,
rlm@432 10 proprioception, muscle tension, and hearing. Each of these senses
rlm@432 11 provides a wealth of parameters that are biologically
rlm@432 12 inspired. =CORTEX= is able to simulate any number of creatures and
rlm@432 13 senses, and provides facilities for easily modeling and creating new
rlm@432 14 creatures. As a research platform it is more complete than any other
rlm@432 15 system currently available.
rlm@432 16
rlm@432 17 In the second half of the thesis I develop a computational model of
rlm@432 18 empathy, using =CORTEX= as a base. Empathy in this context is the
rlm@432 19 ability to observe another creature and infer what sorts of sensations
rlm@432 20 that creature is feeling. My empathy algorithm involves multiple
rlm@432 21 phases. First is free-play, where the creature moves around and gains
rlm@432 22 sensory experience. From this experience I construct a representation
rlm@432 23 of the creature's sensory state space, which I call \phi-space. Using
rlm@432 24 \phi-space, I construct an efficient function for enriching the
rlm@432 25 limited data that comes from observing another creature with a full
rlm@432 26 compliment of imagined sensory data based on previous experience. I
rlm@432 27 can then use the imagined sensory data to recognize what the observed
rlm@432 28 creature is doing and feeling, using straightforward embodied action
rlm@432 29 predicates. This is all demonstrated with using a simple worm-like
rlm@434 30 creature, and recognizing worm-actions in video.
rlm@435 31
rlm@435 32
rlm@435 33 empathy is important
rlm@435 34 cortex tests that idea