rlm@436: Here I demonstrate the power of using embodied artificial intelligence
rlm@436: to attack the /action recognition/ problem, which is the challenge of
rlm@436: recognizing actions performed by a creature given limited data about
rlm@437: the creature's actions, such as a video recording. I solve this
rlm@437: problem in the case of a worm-like creature performing actions such as
rlm@524: curling and wiggling. 
rlm@432: 
rlm@436: To attack the action recognition problem, I developed a computational
rlm@441: model of empathy (=EMPATH=) which allows me to recognize actions using
rlm@441: simple, embodied representations of actions (which require rich
rlm@524: sensory data), even when that sensory data is not actually available.
rlm@551: The missing sense data is imagined by combining previous experiences
rlm@551: gained from unsupervised free play. The worm is a five-segment
rlm@551: creature equipped with touch, proprioception, and muscle tension
rlm@551: senses. It recognizes actions using only proprioception data.
rlm@432: 
rlm@436: In order to build this empathic, action-recognizing system, I created
rlm@436: a program called =CORTEX=, which is a complete platform for embodied
rlm@436: AI research. It provides multiple senses for simulated creatures,
rlm@524: including vision, touch, proprioception, muscle tension, and hearing.
rlm@524: Each of these senses provides a wealth of parameters that are
rlm@436: biologically inspired. =CORTEX= is able to simulate any number of
rlm@436: creatures and senses, and provides facilities for easily modeling and
rlm@436: creating new creatures. As a research platform it is more complete
rlm@436: than any other system currently available.