Multiple local controls integrated by RMPs for FCP-based hexapod walking

Abstract

When hexapod robots walk on uneven terrain, they need to avoid the legs from collisions with obstacles, their own legs, and the ground, while also considering joint limits and reaching their feet to a destination. To address these issues, several local controls are defined in this paper for the swing legs in the Follow-the-Contact-Point (FCP) gait, which are integrated with Riemannian Motion Policies (RMPs). RMPs allow for multiple local control policies to be considered simultaneously for different points on the leg frames. For the local controls, a range sensor is installed on the foot to prevent dragging and contacting the ground firmly, even if the destination is in the air. Additionally, linear position sensors are equipped around the tibia to detect and avoid obstacles. The effectiveness of the proposed method is confirmed in walking experiments on flat and uneven terrain.