Trafficability anticipation for quadruped robot in field operation

Abstract

Abstract The capability to predict catastrophic failures and anticipate challenges relating to trafficability enables quadruped robots to adjust their planned trajectory in time to prevent any damages. In this paper, an estimation‐based method is proposed to describe the validity of the trajectory from the viewpoint of the traction performance of the terrain. The main contribution of this study lies in equipping quadruped robots with the ability to comprehend the strength characteristics of the terrain and assess the validity of trajectory planning by anticipating potential catastrophic motion failures. To this end, a novel perception method of the foot terrain interface based on proprioception is proposed to estimate terrain strength parameters. The vehicle–terrain traction model is extended to a legged locomotion pattern to estimate the traction limits of the terrain. The trafficability anticipation is driven by combining the traction limits of the terrain and the expected ground reaction force, as obtained by the model‐based predictive methods. The validity of this method is verified using a quadruped robot test platform. Results highlight the potential of the presented approach, in which the anticipation of trafficability, grounded in the understanding of the terrain strength, serves as the explicit foundation for altering trajectories to avert motion failure.