Blind hexapod walking over uneven terrain using only local feedback

Abstract

Biological studies on animal walking and running have contributed to the control of mobile robots. Taking inspiration from such studies, a hybrid impulse-position (HIP) controller has been developed for legged locomotion over uneven terrain. The HIP controller generates joint torques during the stance phase of each leg cycle to propel and stabilize the body, and is meant to operate with CPG-based coordination strategies and ground search routines to generate robust walking, running, jumping and climbing. Joint torques are computed without extracting information about the ground and using only local feedback. No feedback of body state is needed to regulate body motion, which is preferred because inertial sensors are rife with issues. The algorithm is applied to a hexapod system in simulation, and an experimental system is being built to further investigate the approach. The HIP controller has potential for expansion to bipeds, quadrupeds and other biologically-inspired forms.