Multi-objective Optimization for Design of Redundant Serial Robots

Abstract

A great number of performance indices have been widely used in design optimization for robots. However, the performance indices used for optimization in previous researches do not evaluate robot performance comprehensively. Therefore, this paper proposes a Global Dexterity Index (GDI) based on condition number and manipulability, and then the kinematic performance indices are divided into 4 categories. Among them, the GDI, the Global Singularity Index (GSI) and the Global Fluctuating Index (GFI) are used as the objective function. In addition, Non-Dominated Sorting Genetic Algorithm (NSGA-II) is used to multi-objective optimization for 7-DOF collaborative robot. Finally, the optimal solution in the Pareto front is selected by comparing the Structure Length Index (SLI). The results indicate that the performance of the optimized robot in this paper is improved significantly compared with the prototype robot.