Particle swarm optimization-based method for navigation of mobile robot in unstructured static and time-varying environments

摘要

The paper considers a problem of the mobile robot navigation in unstructured environments based on an improved Particle Swarm Optimization (PSO) method. It introduces a new fitness function divided into sub-functions, where each sub-function is responsible for accomplishing given objectives and imposed constraints. The main objectives of proposed sub-functions are to maximize the distance between the robot and a previously detected nearest obstacle, minimize distance from the robot to a global best position of the particles and the local minima problem avoidance. Also, proper constraints are included to prevent the robot from penetrating into restricted (prohibited) areas around detected obstacles or to avoid that particle moves too far away from the mobile robot's current position. The proposed PSO based planner generates a smooth collision free path with a low computational cost and approaches the goal position employing an adaptive control strategy. The adjustment and selection of suitable parameters values of PSO method are based on extensive simulations performed in various environments. The effectiveness of the proposed method has been verified with a series of simulations.