Method for tracking constant force surface of robot based on reinforcement learning

Abstract

The invention discloses a method for tracking a constant force surface of a robot based on reinforcement learning. The method comprises the steps that (1) for a feature of an industrial robot end effector contacting a surface profile, a mapping relation of surface normal force and a known sensor coordinate system is established; (2) an explicit force controller is designed for the contact processbetween the robot and a surface; (3) parameters of the controller are optimized by using a reinforcement learning method based on Gaussian model speculation; (4) iterative experiment is carried out until the error between the obtained force and the expected force is within a set range. The method for tracking the constant force surface of the robot based on reinforcement learning has the advantages that the problem is solved that constant tracking force is obtained difficultly in existing surface tracking of the robot, prior experience is not needed, the data utilization rate is high, the convergence speed is fast, and the obtained surface profile track provides an initial reference trajectory for subsequent grinding, polishing and other processing.

Description

technical field [0001] The invention relates to the technical field of robot constant force control, in particular to a method for tracking a robot constant force curved surface based on reinforcement learning. Background technique [0002] The contour of the curved surface can be obtained by using the robot end effector to track the curved surface. The stable contour trajectory can provide the initial trajectory for machining, such as grinding and polishing, and reduce the workload of manual teaching. However, the nonlinearity, time-varying and coupling characteristics of the robot in the motion process will cause uncertain factors; insufficient rigidity of the robot itself (such as reducer, open chain structure) will cause the robot to vibrate; the unknown surface profile requires the robot to adjust its own parameters. To adapt to changes in the external environment, the contact force between the robot and the curved surface is often unstable and fluctuates greatly. [0...

AI Technical Summary

Problems solved by technology

However, the nonlinearity, time-varying and coupling characteristics of the robot in the motion process will cause uncertain factors; insufficient rigidity of the robot itself (such as reducer, open chain structure) will cause the robot to vibrate; the unknown surface profile requires the robot to adjust its own parameters. Adapt to changes in the external environment, so the contact force between the robot and the curved surface is often unstable and fluctuates greatly

Images