Double-robot force/position multivariate data driving method based on reinforcement learning

Abstract

The invention discloses a double-robot force / position multivariate data driving method based on reinforcement learning. A master robot employs an ideal position element control strategy, learns an expected position through a reinforcement learning algorithm, and feeds back an actual position to the expected position, and the purpose is to generate an optimal force when the robot interacts with an environment, so that the position error is minimized; and a slave robot employs a damping PD control strategy suitable for an unknown environment based on the force element control strategy of position deviation of the master robot, learns expected acting force through the reinforcement learning algorithm, and is driven to approach the minimum force of an expected reference point. The master robot and the slave robot respectively learn the expected position and the expected acting force through the reinforcement learning algorithm, and both adopt a proportional differential control rate to set respective differential coefficients (kp) and proportionality coefficients (kd). According to the method, the flexibility of double-machine cooperation can be improved, the parameter optimization problem in force / position control is solved, and large errors in the transient state are avoided.

Description

Technical field [0001] The present invention relates to a multi-robot synergistic control technology, and in particular to a two-machine manpower / bit multivariate data driving method based on an intensive learning. Background technique [0002] With the continuous change of the processing volume of the steel / aluminum, the work environment, some work is only difficult to bear on the single robot, and the synergy between multiple robots is required to be completed. Multi-machine collaborative homework has replaced a single machine and become a research hotspot for building intelligent production lines. Multi-robot systems have the characteristics of data redundancy, robustness, robustness, and better data redundancy, better data redundancy, better system space distribution, better data redundancy, and better data redundancy, robustness. Synergism between multiple robots can be reliably completed complex tasks such as high-precision operations and high-efficiency machining of si...

AI Technical Summary

Problems solved by technology

When the environment is unknown, force control is insufficient to generate the desired strength for uncertainties in the environment

Images