Artificial Intelligence: Presentations
Autonomous agents collaborating in nondeterministic worlds need to react to changing environments and communicate new information effectively. In this work, we enable supportive communication, a key collaboration behavior identified in effective human teams, in autonomous agents. We present a decision-theoretic framework and address inconsistencies in team members' mental models. We evaluate and demonstrate the effectiveness of our proposed framework on decentralized agents collaborating in partially observable environments.
Robot manipulation is one of the most important
scientific challenge encountered in most service robotics applications. Important advancements
in the field have been done during the last decade,
but we are still far from a general-purpose robot housekeeper.
Why? In this talk we present our proposed framework
for designing and evaluating robots
capable of basic manipulation capabilities as well as a set of benchmark tasks with robot experimental results.