Models for Slip Estimation and Soft Terrain Characterization With Multilegged Wheel–Legs

Abstract

Successful operation of off-road mobile robots faces the challenge of mobility hazards posed by soft, deformable terrain, e.g., sand traps. The slip caused by these hazards has a significant impact on tractive efficiency, leading to complete immobilization in extreme circumstances. This paper addresses the interaction between dry frictional soil and the multilegged wheel–leg concept, with the aim of exploiting its enhanced mobility for safe, in situ terrain sensing. The influence of multiple legs and different foot designs on wheel–leg–soil interaction is analyzed by incorporating these aspects to an existing terradynamics model. In addition, new theoretical models are proposed and experimentally validated to relate wheel–leg slip to both motor torque and stick-slip vibrations. These models, which are capable of estimating wheel–leg slip from purely proprioceptive sensors, are then applied in combination with detected wheel–leg sinkage to successfully characterize the load bearing and shear strength properties of different types of deformable soil. The main contribution of this paper enables nongeometric hazard detection based on detected wheel–leg slip and sinkage.