An adaptive genetic algorithm for robot motion planning in 2D complex environments

[Display omitted] In this paper, an adaptive genetic algorithm (GA) for robot motion planning in 2D complex environments is proposed. Since the robot motion planning problem is generally an NP-hard problem, metaheuristics such as GA are proper approaches to solve it. Therefore, a new adaptive method based on GA is proposed to solve this problem. In order to overcome the local-trap problem and avoid premature convergence, a novel selection operator is designed. In our model, in each iteration, if necessary, the selective pressure is updated by using feedback information from the standard deviation of fitness function values. This adaptive model helps the proposed method better maintain the diversity of individuals and escape from the local optima. We experimentally compare the proposed method to three other state-of-the-art GA-based approaches. The experimental results confirm that our proposed algorithm outperforms the related methods in terms of solution quality and finding an optimum path.