Parameterized controller generation for multiple mode behavior

Chaohui Gong, Matthew J. Travers, Hsien-Tang Kao, Howie Choset

Abstract: We derive and demonstrate a new capability for snake robots in which two behaviors-one for locomotion and the other for manipulation-are executed simultaneously on the same robot. This is done in two steps: 1) inverse kinematics via numerical optimization and 2) gait-based locomotion via modal decomposition. The result is an analytical representation of a multiple mode behavior that reduces online execution to simple parameterized control. This representation makes it possible to derive a feedback control law that enables reliable visual servoing using a snake robot while climbing a pole.