Multi-waypoint path planning for autonomous robots remains a challenging optimization problem, as traditional methods often struggle with scalability. Inspired by the network formation of Physarum polycephalum, the Slime Mould-based MultiWaypoint Planner (SMMWP), a novel nature-inspired algorithm, is introduced. The navigation task is formulated as a Travelling Salesman Problem. To address this, the standard Slime Mould Algorithm (SMA) is enhanced through two distinct mechanisms. First, priority-based encoding is utilized to guarantee feasible paths. Second, a flow accumulation memory is implemented. Through this feature, the best path segments are persistently reinforced, mimicking how real slime mould thickens active veins over time. The method was extensively evaluated using standard TSP benchmarks. It is observed that SMMWP significantly outperforms GA, PSO, and the standard SMA, with a reduction in path length of up to 62.87% and a decrease in computation time of 69.93%. Furthermore, the planner was integrated into a ROS2 navigation stack for a simulated agricultural robot. The results demonstrate that safe and near-optimal trajectories are generated in real time, effectively combining bio-inspired memory with practical robotic navigation.
Seghier et al. (Fri,) studied this question.