Gait pattern changing of quadruped robot using pulse-type hardware neural networks

The walking and running control of a human musculoskeletal model using a low-power consumption hardware central pattern generator model

Applying a control system with low energy consumption and low load of motion control to robots, similar to living organisms, is considered to be one of the most important issues in robot development. We have been studying systems that use pulse-type hardware neural networks to control robotic motion with a small number of control signals, as is the case in living organisms. In particular, it has been mimicking the function of the central pattern generator localized in the spinal cord of living organisms to generate motion patterns. In the present article, a new biomimetic control system using pulse-type hardware neural networks for biped gait control is reported.