Ying Huang, Minrui Fei, and Wenju Zhou
Dynamic multi-objective optimization (DMO), DNSGA-II, cluster-ing, robot manipulator, Pareto
In a moving obstacle environment, the manipulator performs the multi-objective trajectory planning operation considering the con- sumed time, trajectory length, and joint jerk. Given the unsatis- factory results of the conventional dynamic non-dominated sorting algorithm (DNSGA-II), we propose a K-means clustering DNSGA-II (KMCDNSGA-II) in which the clustering mechanism is introduced for population improvement. In the optimization process, clustering analysis is conducted to investigate the diversity of each genera- tion. In comparison with the traditional DNSGA-II, the improved KMCDNSGA-II and iKMCDNSGA-II enhance the global search ca- pability and extend the Pareto front distribution in the optimization results. After obtaining the Pareto optimal solution set, we compare and analyse the optimization indexes. Finally, by comparing the conventional and improved algorithms in the simulation operation of the manipulator, the validity and feasibility of the improved algorithm are veriļ¬ed.
Important Links:
Go Back
