Robust execution of aggressive maneuvers for planetary robotics

This paper considers the control of planetary rovers that use mobility strategies incorporating aggressive maneuvers such as jumping or hopping. Aggressive maneuvers are difficult to execute since they involve very precise discrete changes in the set of continuous dynamics governing the system. Robustness to noise in the time of initiation of these changes is an appropriate design criterion for control synthesis, which for a linear system can be formulated as a quasiconcave optimization problem and efficiently solved. This paper extends this synthesis method to a representative nonlinear system, the "climbing robot," and demonstrates the effectiveness of the resulting controller.