Testbeds Used for Exploring Learning from Observation

Darrin C. Bentivegna and Christopher G. Atkeson

This paper describes software and hardware environments that we use in exploring ways to have agents learn from observing humans. The two environments to be described are air hockey and the labyrinth marble maze game. While humans perform in the domain the behavior is observed and recorded. This data can then be used for many research endeavors such as seeding a numericallearning algorithm or as input into discovering primitive/macro actions. For our research focus we are interested in using the captured data to teach an automated agent to operate in the environment.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.