An enhanced beluga whale optimization algorithm incorporating cauchy mutation based on swarm-elite learning mechanism and quadratic interpolation strategy for mobile robot path planning

Abstract Mobile robot path planning is a critical prerequisite for ensuring safe, efficient, and stable operations in fields like smart logistics, industrial automation, and service robotics—poor path quality directly compromises task success rates and equipment safety. However, the conventional Beluga Whale Optimization (BWO) algorithm, when applied to this task, suffers from inherent limitations: susceptibility to local optima due to insufficient global exploration, sluggish convergence in later iterations, and inadequate population diversity after the whale fall phase. To address these challenges, this paper first constructs a comprehensive cost function integrating path length minimization, obstacle collision avoidance, and maximum turning angle constraints, effectively transforming the complex 2D mobile robot path planning problem into a multi-constraint optimization problem. Subsequently, an enhanced BWO variant, the CQGBWO algorithm, is proposed, which synergistically incorporates three key improvement strategies: Cauchy mutation with swarm-elite learning, where the top-ranked individuals are selected as elites to guide candidate updates/mutations so as to improve search direction reliability and reduce the risk of premature convergence, quadratic interpolation, and an elite individual-based genetic strategy. Comparative experiments between CQGBWO and five state-of-the-art algorithms are conducted on the CEC2017 benchmark functions. The results demonstrate that CQGBWO achieves superior convergence accuracy and faster convergence speed compared to the competitors. Furthermore, additional experiments are carried out in four path planning scenarios with different levels of complexity to evaluate the path planning capability of CQGBWO. The results indicate that CQGBWO is able to generate safer paths with lower cost functions than those produced by other competing algorithms.