Abstract
A walking vehicle with the ability to negotiate steep and rough terrain offers the potential to significantly improve the science return from future planetary surface exploration missions. An attractive configuration uses a lightweight walking scout adjunct to the main rover; the scout will have to successfully negotiate a wide range of terrains, and therefore a good understanding of the behaviour of a lightweight rover on loose regolith is essential. This study addresses this mission scenario, and describes a new, comprehensive leg/soil interaction model for light walking vehicles on granular terrain. The study demonstrates that the semi-empirical models used with success in other contexts result in inaccurate leg/soil force predictions, and instead applies ideas from the developing study of granular materials, together with a detailed characterisation of the sinkage process, to propose and validate a combined model of terrain interaction based on an understanding of the physics and micro mechanical processes at the granular level. For a walking vehicle, sinkage behaviour is shown to be complex, dynamic and multiphase. Representative models for each phase are developed based on a combination of theoretical analysis and empirical observation, and validated using a purpose designed and constructed kinematically accurate and fully instrumented test environment, the Single Leg Test Bed. A trade-off study of the effects of varying foot geometry is conducted comparing the predicted behaviour, actual results, features and benefits of several competing foot designs. Use of a bespoke Parametric Tool to investigate the implications of the model is described, as are the results of model sensitivity analysis. The computational efficiency of the algorithm is evaluated and the potential for closed loop vehicle control discussed. Finally, steps required to evaluate the developed algorithms in a full vehicle context are introduced, through simulation of full vehicle performance and construction of a representative rover in a multi body dynamics simulation.