Efficient Path Planning in Obstacle-Dense Grids via Advanced Optimization

Path planning in obstacle-dense grid environments remains a challenging problem, as classical planners often trade solution quality for computational speed, while optimization-based methods may suffer from premature convergence. This paper presents a comparative evaluation of three recent metaheuristic algorithms MGO, RIME, and PLO for discrete grid-based robot path planning, using the Rapidly-exploring Random Tree (RRT) algorithm as a baseline. The main contribution of this work is a unified waypoint-based optimization framework that incorporates obstacle penalties and path smoothness into a single fitness function, enabling a fair and reproducible comparison between classical and metaheuristic planners. Experiments are conducted on a 20×20 grid map with 52 static obstacles and evaluated using path cost, length, number of turns, convergence behavior, and execution time. Results from five independent runs show that all metaheuristic approaches significantly outperform RRT, achieving path cost reductions of up to 25%. Among them, PLO attains the lowest average cost (24.42 ± 0.10) and the best individual solution (24.33), while MGO provides the most balanced performance, producing the smoothest trajectories with the fewest turns (3.2 ± 1.5) and a stable average cost (24.68 ± 0.29). RIME converges rapidly in early iterations but exhibits higher variability and premature stagnation. These findings demonstrate the effectiveness of recent metaheuristics for grid-based navigation and provide clear guidelines for selecting optimization strategies based on path quality, robustness, and convergence requirements