Previewing control humanoid robot gait planning method based on deep reinforcement learning

Abstract

The invention discloses a previewing control humanoid robot gait planning method based on deep reinforcement learning. The method comprises steps of 1) acquiring state information through a sensor assembled on a humanoid robot; 2) improving a current deep reinforcement learning network, and defining totally new states, action vectors and awarding functions; 3) using the defined action vectors to correct output of a previewing controller, calculating the angle of each steering engine of the feet of the humanoid robot and guiding the humanoid robot to walk; and 4) during walking processes of thehumanoid robot, using values of the states, the action vectors and the awarding functions to update the improved deep reinforcement learning network. According to the invention, the walking problem of the humanoid robot under a complex environment can be effectively solved; and a test is carried out on a simulation platform and a real robot to verify the validity of the method.

Description

technical field [0001] The invention relates to the technical field of humanoid robots, in particular to a gait planning method for preview control humanoid robots based on deep reinforcement learning. Background technique [0002] An essential function of a humanoid robot is to walk stably. However, due to the complexity of the structure of the humanoid robot, the strong coupling relationship, and the poor independence of the modules, it is difficult to realize the stable walking function of the humanoid robot. Therefore, the gait control and planning of humanoid robots have become a research hotspot in related fields. Traditional gait control methods can be roughly divided into two categories: methods based on modern control theory and methods based on walking mechanism. However, most of these methods are old and not suitable for the increasingly complex model mechanisms. The recent continuous proposal and innovation of various machine learning methods has also stimulat...

AI Technical Summary

Problems solved by technology

[0004] The present invention mainly studies the gait planning function of humanoid robots walking in complex ground environments. Aiming at the fact that the existing control theory cannot effectively solve the problem of walking in complex environments, a preview control humanoid robot based on deep reinforcement learning is proposed. The gait planning method can effectively solve the walking problem of humanoid robots in complex environments, and has been tested on simulation platforms and physical robots to verify the effectiveness of this method

Images