Quadruped robot motion control self-generation and real object migration method

Abstract

The invention discloses a quadruped robot motion control self-generation and real object migration method. A PPO algorithm is used for optimizing a quadruped robot motion controller, parameters influencing quadruped robot motion control performance are determined, a robust controller is learned to reduce the difference between a simulation model and an actual model, and the migration success rate is increased. Firstly, autonomous generation of a control strategy is realized in a simulation environment, then a motion controller learned in simulation is deployed on an actual quadruped robot, pose information of the quadruped robot is acquired in real time through an inertia measurement unit, and the motion controller predicts joint angles of legs of the quadruped robot and outputs the joint angles to corresponding motors. Finally, autonomous gait emergence is realized. According to the method, the problems that a traditional quadruped robot is difficult in control modeling and poor in environmental applicability, and a control method using deep reinforcement learning mostly stays in a simulation stage are solved, so that a deep reinforcement learning algorithm is applied to motion control of an actual quadruped robot, and rapid autonomous emergence of complex adaptive gaits is realized.

Description

technical field [0001] The invention relates to the field of robot motion control, in particular to a quadruped robot motion control self-generation and object transfer method. Background technique [0002] The ultimate goal of my country's deep space exploration and manned missions to the moon is to establish permanent bases on the surface of the moon and Mars. The construction of extraterrestrial celestial body bases is an unprecedented major project that requires adequate preparations before implementation. The most critical One link is to develop space robots that can adapt to the surface environment of the moon and Mars, paving the way for the establishment of bases and even manned landings. [0003] In the complex alien environment, the inherent "suspension structure" of the legged mobile robot separates the body from the terrain environment, and it can move stably and continuously only by relying on discrete footholds, so that it can move in irregular terrain. It show...

AI Technical Summary

Problems solved by technology

However, such control methods often stay in the simulation stage, and the generated stable gait does not perform well in the actual robot. This is due to the large model difference between the simulated physical system and the actual physical system, and this difference In the actual movement will be gradually enlarged

Images