Abstract
It is well-known that a neural circuit in a spinal cord called central pattern generator (CPG) plays a crucial role to generate stable and adaptive bipedal walking. Although various CPG models have been proposed so far, the essential mechanism has not been clarified: neural connections in CPG and sensory feedback have been designed on ad-hoc basis. To reconsider the design principle of CPG model for bipedal walking, we extended our quadruped CPG, which fully exploits physical interaction yielded from local load sensing, to bipedal walking control, based on the phylogeny process. In this study, we developed a bipedal robot that has load sensors in the plantar area for experimental verification of our bipedal walking model.
