Multi-strategy ensemble wind driven optimization algorithm for robot path planning

Abstract

In this study, a multi-strategy ensemble wind driven optimization (MEWDO) algorithm is proposed and combined with cubic spline interpolation to solve path planning challenges for single and multiple robots. The proposed MEWDO uses a Chebyshev map to initialize air particle populations and increase population diversity. A segmented learning local exploitation strategy is proposed to upgrade the exploitation ability of the algorithm. To enhance the exploration ability of the algorithm, a mutation strategy is introduced that disturbs dimensions one by one, based on the F-distribution with asymmetric characteristics. First, performance comparison experiments were conducted between MEWDO and seven other intelligent algorithms on 16 benchmark test functions. The results showed that MEWDO performed the best. Second, path planning simulation experiments were conducted in three static environments to compare MEWDO with three intelligent algorithms and the artificial potential field method, and MEWDO outperformed the comparison algorithms in terms of the planned shortest path and algorithm stability. In some complex rescue environments, multiple robots are frequently sent to perform tasks from different routes to improve the rescue success rate. For this purpose, MEWDO was used to plan task paths for five robots to test its performance in multi-robot path planning. The results showed that MEWDO finds the best route for all five robots to perform the task in a complex environment.