Event Location: NSH 1305

Abstract: Personal mobile robots will soon be operating and closely interacting with us in human environments. Balancing mobile robots can be effective personal robots as they can be tall enough for eye-level interaction and narrow enough to navigate cluttered environments, and they also have the dynamic capabilities to move with speed and grace comparable to that of humans. The work presented in this thesis enables balancing mobile robots to achieve fast and graceful navigation in human environments while handling disturbances and dynamic obstacles. This work particularly focuses on the ballbot, a human-sized mobile robot that balances on a single ball.

The natural dynamics of balancing mobile robots have to be exploited in order to make them achieve fast and graceful motions. This thesis introduces shape-accelerated balancing systems as a special class of underactuated systems to which balancing mobile robots like the ballbot belong. They have a special property wherein non-zero shape configurations result in accelerations in the position space. This thesis presents a trajectory planning algorithm that plans shape trajectories for shape-accelerated balancing systems, which when tracked will result in optimal tracking of desired position trajectories. It also presents experimental results of the ballbot with arms successfully achieving desired position space motions using body lean motions, arm motions, and combinations of the two, and also handle cases where the arms are artificially constrained.

This thesis presents an integrated motion planning and control framework based on sequential composition, which enables balancing mobile robots to achieve graceful navigation. It presents controllers called motion policies that are designed to achieve fast, graceful motions in small domains of the position space that are collision-free. It introduces the gracefully prepares relationship that guarantees a valid sequential composition of motion policies to produce overall graceful motion. It presents an automatic instantiation procedure that deploys these motion policies to fill a map of the environment, and also a motion planner that plans in the space of these gracefully composable motion policies to achieve desired navigation tasks. This thesis also presents experimental results of the ballbot successfully achieving different navigation tasks while handling disturbances and dynamic obstacles.

Committee:Ralph Hollis, Chair

George Kantor

Howie Choset

Russ Tedrake, Massachusetts Institute of Technology