期刊名称：IAENG International Journal of Computer Science
印刷版ISSN：1819-656X
电子版ISSN：1819-9224
出版年度：2019
卷号：46
期号：1
页码：1-11
出版社：IAENG - International Association of Engineers
摘要：Mobile robots, when navigating in diverse environments, rely on solutions to trajectory generation problems for achieving the best path. One of those solutions is a heuristic method named Particle Swarm Optimization (PSO). In a previous study, by using such method, the mobile robot could find the best route towards the target without collision; moreover, PSO offers the benefits of simplicity, ease of implementation, and few parameters to regulate. However, the original PSO algorithm cannot guarantee the optimal solution. Local optima still occur, especially in complex and dynamic environments, due to premature convergence. This causes mobile robot collisions with obstacles and generates a long path to the target. In the present study, in order to overcome the problem of premature convergence, dynamic PSO (DPSO) was developed by using a dynamic inertia function to set parameters to accelerate convergence and re-initialize particles. The DPSO was analytically compared with two other algorithms, namely the original PSO (OPSO) and the Gaussian PSO (GPSO). Finally, the proposed DPSO is combined with Fuzzy Logic for obtaining the best control of leader-follower system. In the results, the proposed DPSO algorithm produced the optimum solution faster with convergence of less than 150 iterations for static obstacles and 200 iterations for moving obstacles, 4% shorter traveled lengths, 13% more smoothness, fast processing and guaranteed avoidance of collisions, and stable movement in reaching the target. When the proposed DPSO is combined with Fuzzy Logic, it can improve leader-follower performance in terms of trajectory control, time traveled to the target, and times response in several environmental conditions.
