In order for humans and robots to collaborate together fluidly, robots must be able to (1) recognize their human teammate's intentions, and (2) automatically adapt to those intentions in an intelligent manner. This thesis makes progress in these areas by proposing a framework that solves these two problems (task-level intent recognition and robotic adaptation) concurrently and holistically, using a single model and set of algorithms for both. The result is a mixed-initiative human-robot interaction that achieves the team's goals.
We extend this framework by additionally maintaining a probabilistic belief over the human's intentions. This allows the robot to continuously assess the risk associated with plan execution, and thereby select adaptations that are safe enough, ask uncertainty-reducing questions when appropriate, and provide a proactive early warning of likely failure.
Thesis Committee: Brian Williams, Leslie Kaelbling, Julie Shah, Patrick Winston, Andreas Hofmann