In this talk, I will present my work on learning policies for navigation. I will talk about three projects. First, I will show how insights from classical mapping and planning can be operationalized in context of learning based policies to arrive at policies that can effectively leverage statistical regularities in layout of real world environments. Second, I will show how such learned policies can be made robust to actuation noise and environment changes through use of visual fixes. Finally, I will describe some preliminary work on how we can learn policies for task-agnostic exploration of novel environments.
(video of this talk is upon request)
October 31 @ 12:30
12:30 pm (1h)
Discovery Building, Orchard View Room