Environment Perception and Motion Strategy for Transformable Legged Wheel Robot on rough terrains

Abstract

The transformable legged wheel robot is capable of transforming between leg mode and wheel mode. The leg mode has better overcoming obstacle ability, whereas the wheel mode has swift maneuverability on continuous terrain. This paper presents a robust perception system and an obstacle crossing strategy for a three-mode transformable legged wheel robot in rough terrain. The proposed perception system applies different sensors to acquire the exact terrain information as well as the status of the robotic system. Binocular stereo vision is applied to generate the disparity map and extract the terrain features. The IMU and the current sensors estimate and supervise the real-time safety of the robotic system. Experiment validations of the perception system and obstacle detection are conducted in the field, which indicates that the perception system can provide both the obstacle crossing guidance and the robotic system protection for the robotic system.