Fusion of improved RRT and ant colony optimization for robot path planning

Abstract To address the issues of poor guidance at the beginning of the Ant Colony Optimization (ACO) algorithm, non-smooth paths, and its tendency to fall into local optima, this paper proposes a path planning approach based on the Rapidly-exploring Random Tree (RRT) and Ant Colony Optimization (ACO). Firstly, obstacles are inflated to set a safety distance, and a differentiated pheromone distribution is created using the sub-optimal trajectory produced by the improved RRT, guiding the initial direction of the ant colony. Secondly, dynamic strategies are introduced into the evaporation coefficient and heuristic factor, adjusting their weights according to the number of iterations to enhance the attraction of the target point to the ants. Then, a reward-punishment mechanism is used to update the pheromone, solving the problem of local optima. Finally, a pruning optimization strategy based on the maximum turning angle is employed to remove redundant nodes, making the path smoother. Multiple simulation results confirm that the algorithm possesses good global search capabilities and robustness under various conditions.