Mobile Robot Localization Based on the Novel Leader-Based Bat Algorithm

This work proposes a new approach to the well-known method bat algorithm for solving the mobile robots global localization problem. The proposed method is leader-based bat algorithm (LBBA). The LBBA uses a small number of better micro-bats as leaders to influence the colony in the search for the best position, dealing satisfactorily with ambiguities during the localization process. The tests covered different scenarios aiming at comparing the proposed algorithm with other methods, such as the standard BA, the particle swarm optimization and particle filter. The results outperformed the compared methods, presenting a fast response and errors below the intended tolerance. The algorithm was tested in the robot kidnapping scenario and shows fast recovery in both simulation and in a real environment. In addition, the proposed technique showed 21% lower average error when compared with an algorithm that presents a variable quantity of particles, i.e. the adaptive Monte Carlo localization algorithm.