Automated parameter tuning of a Generic Rover Dynamics Model for autonomous planetary space capability development, verification and validation of space applications

A Generic Rover Dynamic Model (GRDM) is utilised to simulate a rover´s ability to traverse an unstructured environment, such as a Martian terrain, including slippage on sand and rock climbing ability. This model has been developed by Airbus Defence and Space Ltd. within the scope of a UK Space Agency project (CREST-2) to support the study of rover Guidance, Navigation and Control (GNC) and operations as well as for preparation, rehearsal and replay of field trials. The model was developed to run much faster than real time by considering a simplified wheel-terrain interaction model. Hence, the wheel-terrain interaction configuration parameters have to be tuned to obtain good correlation with empirical data. In order to automate the tuning procedure, optimisation methods are applied. The optimisation procedure tunes the generic rover dynamic model w.r.t both sand traversing and rock climbing empirical data. The former consists of a typical continuous cost function whereas the latter fall under the class of satisfiability problems. Due to the differing nature of the sand and rock scenarios, a three step procedure is adopted. In this paper, a three step optimisation procedure is implemented in MATLAB and results are presented.