パルス形ハードウェアニューラルネットワークによる筋骨格モデルの2足歩行制御

抄録

Biped control method of the musculoskeletal model by hardware neural networks using CMOS circuit that does not depend on conventional digital control is reported. By using electronic circuits, the development of robotic systems that are similar to the nervous system of living organisms can be expected. Humans have many muscles in our body, but we do not control them independently. Humans have many muscles in their bodies. However, they are not controlled independently of each other. Humans operate by coordinating the muscles necessary for leg motion through control signals generated by the central pattern generator localized in the spinal cord. We proposed a CPG model that uses a pulse-type hardware neural networks to generate signals to control human walking and running. Using the proposed CPG model, the gait control method of the musculoskeletal model was proposed.