An Attitude Motion Planning Algorithm for One-legged Hopping Robot Based on Spline Approximation and Particle Swarm Optimization

摘要

The optimal control inputs obtained by the traditional robot motion optimization methods are nonzero at the start and end of robot motion, making it difficult to control the attitude motion of the robot directly with the motor. To solve the problem, this paper combines the particle swarm optimization (PSO) with spline approximation into a new method, which can replace the Fourier approximation in traditional algorithm. Firstly, the author set up the dynamic model of the system, and transformed the motion planning problem with nonholonomic constraints into the optimal control problem, under the conservation of angular momentum. Next, the spline approximation and the PSO were employed to optimize the trajectory of the attitude motion of the system, and control the inputs to zero at the start and end of robot motion, such that the robot could move from the initial position to the desired destination in a motion cycle. The proposed algorithm was proved through numerical experiment as capable of effectively controlling the attitude motion of nonholonomic hopping robot.