Robot intelligent self-adaptive compliance control method under unknown environment

Abstract

The invention provides a robot intelligent self-adaptive compliance control method under an unknown environment. According to the method, particle swarm operation parameters are optimized via image knowledge transfer learning, parameter identification is conducted to an environmental contact mechanical model based on an optimized particle swarm identification algorithm, and self-correction adjustment of an impedance control reference position is guided based on the parameter identification result to realize a compliance control effect of accurate force position dual-control under the unknown environment. The method combines the particle swarm identification algorithm based on the contact mechanical model and automatic-correction adjustment of the impedance control reference position, the unknown mechanical properties of linear elastic and non-linear elastic materials, accurate force position synergistic control is realized; the transfer learning optimization method of the particle swarm operation parameters is proposed, algorithm operation parameter identification knowledge is learned from historical identification tasks, the optimized particle swarm identification algorithm operation parameters are obtained based on characteristics of target identification tasks, and the calculation efficiency of a parameter identification system under the unknown environment is improved.

Description

technical field [0001] The invention belongs to the field of force-position dual control of robots, and in particular relates to an intelligent self-adaptive compliant control method for robots in unknown environments. Background technique [0002] Robots have been more and more widely used in industry and service industries, and the tasks they face no longer only require position control. When facing many workplaces such as robot grinding, assembly operations, human massage, and rehabilitation operations, Dual control of robot position and force such as impedance control is required. The above operations often face the problem of unknown working environment: on the one hand, the operation object has both linear elastic characteristics and compliant materials with nonlinear elastic characteristics; There are changes due to physical signs, organizational structure and other factors, and the unknown working environment often leads to large error in impedance control force tra...

AI Technical Summary

Problems solved by technology

The above operations often face the problem of unknown working environment: on the one hand, the operation object has both linear elastic characteristics and compliant materials with nonlinear elastic characteristics; There are changes due to factors such as physical signs and organizational structure, and the unknown working environment often leads to large error in impedance control force tracking or even loss of force tracking ability

However, the neural network often needs a lot of offline training. The genetic algorithm solution process in the evolutionary algorithm involves tedious encoding and decoding processes, and the calculation amount is relatively large. Although the particle swarm optimization algorithm is relatively simple in the solution process and calculation amount, due to the operation Due to factors such as parallel computing characteristics, the calculation amount of the particle swarm algorithm is still greater than that of the least squares algorithm, which will have a negative impact on the calculation efficiency of the system

Images