Trajectory planning method, device, storage medium and program product of mechanical arm

Abstract

The present application provides a trajectory planning method, device, storage medium and program product for a robotic arm. The method includes: determining the initial state and state transition function of a target system when the target system executes a target task; applying a differential dynamic programming algorithm to the initial state and the state transition function to obtain an initial trajectory when controlling the target system to execute the target task through the robotic arm; and obtaining a final trajectory when controlling the target system to execute the target task through the robotic arm according to the initial trajectory and a model-free reinforcement learning method. Thus, relatively accurate trajectory planning can be performed on the robotic arm.

Images