Optimal Periodic Hopping Trajectory Generation for Legged Robots

Abstract

In this paper, we describe an optimal periodic hopping trajectory generation method for legged robots and verified its performance through experiments. Important state variables, namely initial position, initial velocity, and take-off velocity, are defined for the periodic hopping motion and optimization is used to find them. The objective function used for the optimization is composed of three variables which the hopping robot should consider: maximum power of the knee, the root-mean-square (RMS) torque of the knee, and impact force. Each variable affects the trajectory according to weight, which was confirmed by simulation. The 2-link robot used in the experiment is called PONGBOT -LEG, which is a monopod robot based on high speed and high torque motors. Furthermore, the results from the experiments were analyzed and showed both similarities and differences with the simulations. Based on this research, we aim to develop a quadruped robot capable of flying trotting.