diff thesis/abstract.org @ 524:8e52a2802821

incorporating winston's changes.
author Robert McIntyre <rlm@mit.edu>
date Sun, 20 Apr 2014 21:46:46 -0400
parents c20de2267d39
children d304b2ea7c58
line wrap: on
line diff
     1.1 --- a/thesis/abstract.org	Thu Apr 03 20:48:09 2014 -0400
     1.2 +++ b/thesis/abstract.org	Sun Apr 20 21:46:46 2014 -0400
     1.3 @@ -3,20 +3,22 @@
     1.4  recognizing actions performed by a creature given limited data about
     1.5  the creature's actions, such as a video recording. I solve this
     1.6  problem in the case of a worm-like creature performing actions such as
     1.7 -curling and wiggling.
     1.8 +curling and wiggling. 
     1.9  
    1.10  To attack the action recognition problem, I developed a computational
    1.11  model of empathy (=EMPATH=) which allows me to recognize actions using
    1.12  simple, embodied representations of actions (which require rich
    1.13 -sensory data), even when that sensory data is not actually
    1.14 -available. The missing sense data is ``imagined'' by the system by
    1.15 -combining previous experiences gained from unsupervised free play.
    1.16 +sensory data), even when that sensory data is not actually available.
    1.17 +The missing sense data is ``imagined'' by the system by combining
    1.18 +previous experiences gained from unsupervised free play. The worm is a
    1.19 +five-segment creature equipped with touch, proprioception, and muscle
    1.20 +tension senses. It recognizes actions using only proprioception data.
    1.21  
    1.22  In order to build this empathic, action-recognizing system, I created
    1.23  a program called =CORTEX=, which is a complete platform for embodied
    1.24  AI research. It provides multiple senses for simulated creatures,
    1.25 -including vision, touch, proprioception, muscle tension, and
    1.26 -hearing. Each of these senses provides a wealth of parameters that are
    1.27 +including vision, touch, proprioception, muscle tension, and hearing.
    1.28 +Each of these senses provides a wealth of parameters that are
    1.29  biologically inspired. =CORTEX= is able to simulate any number of
    1.30  creatures and senses, and provides facilities for easily modeling and
    1.31  creating new creatures. As a research platform it is more complete