Topological Progress Potential-Enhanced Continuous-Space Ant Colony Algorithm for Robot Path Planning

To address the issues of traditional grid-based Ant Colony Optimization path planning in discretized continuous space-including limited direction freedom, lack of global topological guidance, and difficulty in balancing path smoothness and safety margin-a topological progress potential-enhanced continuous-space ant colony path planning algorithm (TPP-CSACO) is proposed. TPP-CSACO discards grid-based expansion; instead, a perception circle centered on each ant is defined, movement is executed via a sector-based perception framework with probabilistic direction selection, and band-shaped decaying pheromones are deposited along the path. By coupling the global topological progress potential derived from the simplified probabilistic roadmap (PRM) with pheromones, a dual-field guidance mechanism is established to prevent local congestion. Combined with the explicit safety constraints of the signed distance field (SDF), an adaptive step size strategy that integrates elastic step size and frustration-induced temperature rise is introduced to enhance obstacle avoidance and search stability. Results from repeated experiments on multiscale constrained maps (conducted against six typical algorithms and the traditional ACO) show that compared with ACO, TPP-CSACO reduces the path length by up to 50.6% in the same environment, while achieving faster convergence and maintaining good search diversity. Although the path length increases slightly (by a maximum of 5.9%) compared with the shortest heuristic algorithms, the maximum turning angle is reduced by 75% to 93%, and a 100% success rate and zero safety violations are realized. This indicates that TPP-CSACO has achieved a relatively stable balance among safety, smoothness, and global search capability.