Adaptive differential evolution for collaborative path planning of multiple unmanned aerial vehicles

This paper proposes an adaptive differential evolution combined with dynamic Thompson sampling, called DE-DYTS, to solve the collaborative path planning problem of multiple unmanned aerial vehicles (UAVs). Compared to single UAV path planning, the coordination time and collision avoidance among UAVs are considered in this study. Dual-layer encoding is designed to reduce the number of path control points. Four differential evolution operators are used to construct a pool of candidate operators to improve the optimization efficiency. Dynamic Thompson sampling is employed to learn the expected returns of the operators. DE-DYTS is tested in three scenarios to evaluate its effectiveness and efficiency, and the results are statistically analyzed using the Wilcoxon rank sum test. Experimental results show that DE-DYTS achieves significant superiority over advanced variants of differential evolution, especially in scenarios with a large number of UAVs.