Data Fusion Applied to the Leader-Based Bat Algorithm to Improve the Localization of Mobile Robots

Abstract

The increasing demand for autonomous mobile robots in complex environments calls for efficient path-planning algorithms. Bio-inspired algorithms effectively address intricate optimization challenges, but their computational cost increases with the number of particles, which is great when implementing algorithms of high accuracy. To address such topics, this paper explores the application of the leader-based bat algorithm (LBBA), an enhancement of the traditional bat algorithm (BA). By dynamically incorporating robot orientation as a guiding factor in swarm distribution, LBBA improves mobile robot localization. A digital compass provides precise orientation feedback, promoting better particle distribution, thus reducing computational overhead. Experiments were conducted using a mobile robot in controlled environments containing obstacles distributed in diverse configurations. Comparative studies with leading algorithms, such as Manta Ray Foraging Optimization (MRFO) and Black Widow Optimization (BWO), highlighted the proposed algorithm's ability to achieve greater path accuracy and faster convergence, even when using fewer particles. The algorithm consistently demonstrated robustness in bypassing local minima, a notable limitation of conventional bio-inspired approaches. Therefore, the proposed algorithm is a promising solution for real-time localization in resource-constrained environments, enhancing the accuracy and efficiency in the guidance of mobile robots, thus highlighting its potential for broader adoption in mobile robotics.