Particle swarm optimization for solving a scan-matching problem based on the normal distributions transform

In this paper, an evolutionary scan-matching approach is proposed to solve optimization issue in simultaneous localization and mapping (SLAM). A rich literature has been invested direction, however, most of the approaches lack fast convergence simplicity regarding process, which should directly affect accuracy environment’s map estimated pose. It a line research that always active, offering various solutions issue. Among many SLAM methods, normal distributions transform (NDT) shown high performances, where numerous works have published up date studies demonstrate its efficiency wrt other methods. Nevertheless, few interested The solution based on NDT using particle swarm (PSO) it dubbed NDT-PSO. main contribution pose estimation problem PSO iterative maps. performances NDT-PSO proven real experiments performed car-like mobile robot both static dynamic environments. tested for different sizes, results show 70 particles are more than enough find best while avoiding local minima even loop closing. algorithm also suitable time applications, with average runnnig $$145 \rm{ms}$$ iterations process. This value can be further reduced fewer iterations. evaluated methods widely used among operating system community outperforms these algorithms.