Method for improving position and attitude accuracy of industrial robots based on multi-objective optimization

The accuracy requirements of industrial robots are increasingly higher. Firstly, to improve the accuracy performance of industrial robots, this article proposes a full pose kinematic error model based on Modified Denavit-Hartenberg (M-DH), which can better describe the error of industrial robots. Secondly, this article constructs the error fitness function of position and attitude respectively. The multiple objective particle swarm optimization (MOPSO) algorithm is combined to achieve accurate identification of kinematic parameters. Thus, the problems of the position error and attitude error with different scales and magnitudes are solved. Finally, the effectiveness of the MOPSO algorithm is evaluated through experiments. The experimental results show that the average position error and the average attitude error of the Staubli TX60 industrial robot are reduced by 81. 04% and 66. 64%, respectively. Compared with the single-objective optimization algorithms based on the Levenberg-Marquardt (LM) algorithm, the Beetle antennae search swarm optimization (BSO) algorithm, and the particle swarm optimization (PSO) algorithm, the MOPSO optimization algorithm presented in this article is the best method in terms of the generalization in kinematic parameters identification and the maximum pose error optimization. © 2023 Science Press. All rights reserved.