Hybrid fastslam approach using genetic algorithm and particle swarm optimization for robotic path planning

Abstract

Simultaneous Localization and Mapping (SLAM) is an algorithmic technique being used for mobile robot to build and create a relative map in an unknown environment. FastSLAM is one of the SLAM algorithms, which is capable of speeding up convergence in robot’s path planning and environment map estimation. Besides, it is popular for its higher accuracy compared to other SLAM algorithms. However, the FastSLAM algorithm suffers from inconsistent results due to particle depletion problem over time. This research study aims to minimize the inconsistency in FastSLAM algorithm using two soft computing techniques, which are particle swarm optimization (PSO) and genetic algorithm (GA). To achieve this goal, a new hybrid approach based on the mentioned soft computing techniques is developed and integrated into the FastSLAM algorithm to improve its consistency. GA is used to optimize particle weight while PSO is used to optimize the robot’s estimation in generating an environment map to minimize particle depletion in FastSLAM algorithm. The performance of the proposed hybrid approach is evaluated using root mean square error (RMSE) analysis to measure degree of error during estimation of robot and landmark position. The results are verified using margin error analysis. With the percentage error analysis results, the new hybrid approach is able to minimize the problems in FastSLAM algorithm and managed to reduce the errors up to 33.373% for robot position and 27.482% for landmark set position.