Affiliations: [a] School of Computer Science, Wuhan University, Wuhan, China
| [b] School of Computer Science, South-Central University for Nationalities, Wuhan, China
| [c] Hubei Provincial Engineering Research Center for Intelligent Management of Manufacturing Enterprises, South-Central University for Nationalities, Wuhan, China
| [d] School of Information Engineering, Wuhan Technology and Business University, Wuhan, China
| [e]
Southwest Institute of Technology and Physics, Cheng Du, China
Correspondence:
[*]
Corresponding author. Yuanxiang Li, School of Computer Science, Wuhan University, Wuhan, China. E-mail: [email protected].
Abstract: To overcome the disadvantages of low optimization accuracy and prematurity of the canonical PSO algorithm, we proposed an improved particle swarm optimization based on the interaction mechanism between leaders and individuals (PSO-IBLI), and used it to implement robot global path planning. In the PSO-IBLI algorithm, in different stages, each particle learns from the elites according to different regular. Moreover, the improved algorithm divides the execution state into two categories, where the parameters and the evaluation mechanisms are varied accordingly. In this way, the global best particles no longer walk randomly and have more learning objects. At the same time, other particles learn from not only the global best position, their historical best positions, but also the other elites. The learning strategy makes the search mode always in the adaptive adjustment, and it improves the speed of convergence and promotes this algorithm to find a more precise solution. The experimental results suggest that the precision and convergence speed of the PSO-IBLI algorithm is higher than the other three different algorithms. Additionally, some experiments are carried out to plan the robot’s entire collision-free path using the PSO-IBLI algorithm and the other three algorithms. The results show that the PSO-IBLI algorithm can obtain the shortest collision-free way in four algorithms.
