Abstract
To address the limitations of mobile robots in path planning within farmland-specific environments, this paper proposes a biomimetic model: Multi-strategy Collaborative Evolution Honey Badger Algorithm (MCEHBA), MCEHBA achieves improvements through the following strategies: firstly, integrating a sinusoidal function-based nonlinear convergence factor to dynamically balance global exploration and local exploitation; secondly, combining the differential evolution strategy to enhance population diversity, and utilizing gravity-centred opposition-based learning to improve solution space search efficiency; finally, constructing good point set initialization and decentralized boundary constraint handling strategy to further increase convergence accuracy and speed. This paper validates the effectiveness of the optimization strategy and the performance of MCEHBA through the CEC2017 benchmark function set, and assesses the statistical significance of the results using the Friedman test and Nemenyi test. The findings demonstrate that MCEHBA exhibits excellent optimization capabilities. Additionally, this study applied MCEHBA to solve three engineering application problems and compared its results with six other algorithms, showing that MCEHBA achieved the minimum objective function values in all three cases. Finally, simulation experiments were conducted in three farmland scenarios of varying scales, with comparative tests against three state-of-the-art algorithms. The results indicate that MCEHBA generates paths with minimized total costs, demonstrating superior global convergence and engineering applicability.