Multi-Robot Map Fusion Using an N-Mutation Particle Swarm Optimization Algorithm

Aiming at the map fusion problem in multi-robot collaborative Simultaneous Localization and Mapping (SLAM) systems, this paper proposes a map fusion method based on an N-Mutation Particle Swarm Optimization (N-Mutation PSO) algorithm. By analyzing the dissimilarity of grid maps and the alignment features in overlapping regions, a fusion fitness function is designed. Leveraging the global search capability of the PSO algorithm, the optimal coordinate transformation matrix is solved. To address the shortcoming of traditional PSO algorithms stagnating at local minima, a particle age structure and a mutation mechanism are introduced: when a particle's personal best position remains unchanged for a sustained period, a random position-velocity mutation is triggered to escape local optima, and the balance between exploration and exploitation capabilities is maintained by dynamically adjusting the number of mutation operations. Optimization experiments using CEC2017 datasets (IEEE Congress on Evolutionary Computation 2017) functions demonstrate that the proposed algorithm achieves improvements in both convergence accuracy and convergence speed compared to the standard PSO algorithm. Results from map fusion experiments validate the effectiveness of the proposed algorithm for the multi-robot map fusion problem.