Rough-terrain mobile robot planning and control with application to planetary exploration

Abstract

Future planetary exploration missions will require mobile robots to perform difficult tasks in highly challenging terrain, with limited human supervision. Current motion planning and control algorithms are not well suited to rough-terrain mobility, since they generally do not consider the physical characteristics of the rover and its environment. Failure to understand these characteristics could lead to rover entrapment and mission failure. In this thesis, methods are presented for improved rough-terrain mobile robot mobility, which exploit fundamental physical models of the rover and terrain. Wheel-terrain interaction has been shown to be critical to rough terrain mobility. A wheel-terrain interaction model is presented, and a method for on-line estimation of important model parameters is proposed. The local terrain profile also strongly influences robot mobility. A method for on-line estimation of wheel-terrain contact angles is presented. Simulation and experimental results show that wheel-terrain model parameters and contact angles can be estimated on-line with good accuracy. Two rough-terrain planning algorithms are introduced. First, a motion planning algorithm is presented that is computationally efficient and considers uncertainty in rover sensing and localization. Next, an algorithm for geometrically reconfiguring the rover kinematic structure to optimize tipover stability margin is presented. Both methods utilize models developed earlier in the thesis.

(cont.) Simulation and experimental results on the Jet Propulsion Laboratory Sample Return Rover show that the algorithms allow highly stable, semi-autonomous mobility in rough terrain. Finally, a rough-terrain control algorithm is presented that exploits the actuator redundancy found in multi-wheeled mobile robots to improve ground traction and reduce power consumption. The algorithm uses models developed earlier in the thesis. Simulation and experimental results show that the algorithm leads to improved wheel thrust and thus increased mobility in rough terrain.