Research on EPSQP Algorithm for Motion Planning of Humanoid Robot

Abstract

Aiming at the instability and low convergence rate caused by complex dynamics, linear constraint and sensitivity to initial configuration state during Sequential Quadratic Programming (SQP) of motion control of humanoid robot solved by parameter optimization method, firstly, establish a general dynamics model for 7 connecting rods of humanoid robot and establish a simplified, multilevel and inverted pendulum model which falls to the ground from the front to the back for falling motion; secondly, introduce a parameter optimization technique to the optimum control of humanoid robot motion, build the optimum control model, adopt a filter for initial configuration state, improve SQP filter algorithm and enhancing technique, propose Enhancing Parametric Sequential Quadratic Programming (EPSQP) algorithm to improve optimizing process and accelerate convergence rate; finally, verify the validity and optimality of EPSQP algorithm proposed in this paper by computer simulation experiment.