A Multi‐Strategy Fusion for Mobile Robot Path Planning via Dung Beetle Optimization

ABSTRACT In recent years, robot path planning has become a critical aspect of autonomous navigation, especially in dynamic and complex environments where robots must operate efficiently and safely. One of the primary challenges in this domain is achieving high convergence efficiency while avoiding local optimal solutions, which can hinder the robot's ability to find the best possible path. Additionally, ensuring that the robot follows a path with minimal turns and reduced path length is essential for enhancing operational efficiency and reducing energy consumption. These challenges become even more pronounced in high‐dimensional optimization tasks where the search space is vast and difficult to navigate. In this article, a multi‐strategy fusion enhanced dung beetle optimization algorithm (MIDBO) is introduced to tackle key challenges in robot path planning, such as slow convergence and the problem of local optima, and so on, in which MIDBO incorporates several key innovations to enhance performance and robustness. First, the Tent chaotic strategy is used to diversify initial solutions during population initialization, thereby mitigating the risk of local optima and improving global search capability. Second, a penalty term is integrated into the fitness function to penalize excessive turning angles, aiming to reduce the frequency and magnitude of turns. This modification results in smoother and more efficient paths with reduced lengths. Third, the inertia weight is adaptively updated by a sine‐based mechanism, which dynamically balances exploration and exploitation, accelerates convergence, and enhances algorithm stability. To further improve efficiency for path planning, the MIDBO integrates a Levy flight strategy and a local search mechanism to boost the search capability during the stealing phase, contributing to smoother and more practical paths planned for the robot. A series of thorough and reproducible experiments are performed using benchmark test functions to evaluate the performance of MIDBO in comparison to several leading metaheuristic algorithms. The results demonstrate that MIDBO achieves superior outcomes in path planning tasks with optimal and mean path lengths of 42.1068 and 44.4755, respectively, which significantly outperforms other algorithms including IPSO (47.6244, 55.9375), original DBO (47.6244, 55.9375), and ISSA (47.6244, 55.9375). MIDBO also markedly reduces the number of turns by achieving best and average values of 10 and 13.4, respectively, compared with IPSO (11, 16.1), original DBO (12, 15.3), and ISSA (12, 16.4). Besides, the consistent performance of MIDBO is confirmed via stability analysis based on the mean square error of path lengths and turn counts across 10 independent trials. For the high‐dimensional optimization tasks, MIDBO achieves 8 and 7 functions about top rankings on 50‐ and 100‐dimensional functions, and specifically MIDBO outperforms DBO, IPSO, and ISSA on 13, 18, and 11 functions, respectively. Therefore, the findings validate MIDBO is a competitive solution of path planning for mobile robot navigation with complex requirements.