Enhancing jumping performance of bionic leg with pneumatic muscle actuation based on neural network control

Most robot bionic legs are driven by motors, but bionic legs driven by pneumatic artificial muscles (PAM) have the advantages of being lightweight with better bionics and flexibility. In this paper, a new bionic jumping leg driven by PAMs was designed and its jumping performance studied. To address the nonlinearity of the PAM and the joint disturbance of the multi-joint bionic leg, a neural network PID controller was designed to achieve nonlinearity and better dynamic performance. The joint position was planned based on changes in the mass center of the bionic leg. Subsequently, RBF-PID and BP-PID controllers were utilized for joint position control simulation and jumping experiments. Both RBF-PID and BP-PID control achieved a response time of 0.01 s. BP-PID control demonstrated superior position control accuracy over RBF-PID control, and the bionic leg achieved a jumping height of 220 mm. In summary, the new neural network PID controller outperforms the traditional PID controller in the terms of response speed and control accuracy. BP-PID is better than RBF-PID control in terms of position control accuracy, and the leg can reach a jumping height of 220 mm.