Pattern generation and control of a double pendulum using a nonlinear oscillator

Prior to developing the wearable walking assistant robot that supports walking for hemiplegia patients, a neural oscillator, a type of CPG (central pattern generation), was applied to 2-DOF double pendulum, which can replace the leg of a robot. The walking pattern generation method was proposed using the gait pattern of the non-affected side of hemiplegia patients. Because it is difficult for hemiplegia patients to distinguish the intended action signal of the patient wearing a robot on their affected side, we had to utilize a limited amount of information to get the maximum effect. We needed an effective solution for robot control in an outside environment where many unknown disturbances exist. In order to deal with these two problems, we used a nonlinear oscillator with a double pendulum as a test bed to explore the possibility of producing a robot walking pattern.