Predictable Mobility: A Statistical Approach for Planetary Surface Exploration Rovers in Uncertain Terrain

Abstract

In this article, a statistical mobility prediction for planetary surface exploration rovers has been described. This method explicitly considers uncertainty of the terrain physical parameters via SRSM and employs models of both vehicle dynamics and wheel-terrain interaction mechanics. The simulation results of mobility prediction using three different techniques, SMC, LHSMC, and SRSM, confirms that SRSM significantly improves the computational efficiency compared with those conventional methods. The usefulness and validity of the proposed method has been confirmed through experimental studies of the slope traversal scenario in two different terrains. The results show that the predicted motion path with confidence ellipses can be used as a probabilistic reachability metric of the rover position. Also, for the slope-traversal case, terrain parameter uncertainty has a larger influence on the lateral motion of the rover than on longitudinal motion. Future directions of this study will apply the proposed technique to the path-planning problem. Here, confidence ellipses will be used to define collision-free areas, which will provide useful criteria for generating safe trajectories.