A composite objective function specifically tuned for multi-robot path planning

Authors : Bilal Özak, Mustafa Gül

Pages : 81-88

View : 53 | Download : 163

Publication Date : 2025-12-25

Article Type : Research Paper

Abstract :In swarm robotic systems, the problem of multi-robot path planning (MRPP), especially in obstacle-filled environments, presents significant challenges in terms of coordinated navigation. This work proposes a new fitness function for online MRPP in environments containing dynamic obstacles. The proposed method optimizes the collision-free paths of 20 robots using metaheuristic algorithms such as Particle Swarm Optimization (PSO), Artificial Bee Colony Optimization (ABC), Ant Colony Optimization (ACO), Genetic Algorithm (GA), and Differential Evolution (DE). The fitness function balances conflicting objectives such as proximity to the target, obstacle, and collision avoidance with other robots. Simulations have confirmed that 20 robots divided into two groups safely reach their destinations in a 100x100 unit environment containing dynamic and static obstacles. Simulations showed that the ABC algorithm achieved the best average path length (2667.01 units) in static environments, while PSO provided the fastest computation time (13.15 s). In dynamic environments, ABC again outperformed others in path length (2790.13 units), and PSO remained the fastest (17.30 s). The contributions of the work are a new fitness function, a path planning framework that improves the efficiency of metaheuristic algorithms, and demonstration of the success of the method in dynamic environments.
Keywords : Sürü robotlar, Yol planlama, Çarpışmadan kaçınma, Meta-sezgisel algoritmalar

ORIGINAL ARTICLE URL