In this talk I will present the results of a project in which our research team taught a robot how to perform a complex task without previous knowledge of action primitives or the hierachies that made up the complex task. I will then discuss what we learned also in this project about unforeseen challenges that remain in teaching robots to do tasks.
Candy Sidner is a research professor in Computer Science at Worcester Polytechnic Institute. She is a fellow of the AAAI, an ACL fellow, and past president of the ACL. She has sat on a number of journal editorial boards, been conference chair of NA-ACL, IUI, program chair of IUI, ACL and SigDial, as well as sitting on numerous scientific boards. Her interests span human communication, collaboration, agent and robot interfaces and social behavior in human-machine interaction.