Enhancing FastSLAM 2.0 performance using a DE Algorithm with Multi-mutation Strategies

Authors

Hadeer Adel Attia, Mohamed Arafa, Mohamed TALAAT FAHIM

Abstract

FastSLAM 2.0 is considered one of the popular approaches that utilizes a Rao-Blackwellized particle filter for solving simultaneous localization and mapping (SLAM) problems. It is computationally efficient, robust and can be used to handle large and complex environments. However, the conventional FastSLAM 2.0 algorithm is known to degenerate over time in terms of accuracy because of the particle depletion problem that arises in the resampling phase. In this work, we introduce an enhanced variant of the FastSLAM 2.0 algorithm based on an enhanced differential evolution (DE) algorithm with multi-mutation strategies to improve its performance and reduce the effect of the particle depletion problem. The Enhanced DE algorithm is used to optimize the particle weights and conserve diversity among particles. A comparison has been made with other two common algorithms to evaluate the performance of the proposed algorithm in estimating the robot and landmarks positions for a SLAM problem. Results are accomplished in terms of accuracy represented by the positioning errors of robot and landmark positions as well as their root mean square errors. All results show that the proposed algorithm is more accurate than the other compared algorithms in estimating the robot and landmark positions for all the considered cases. It can reduce the effect of the particle depletion problem and improve the performance of the FastSLAM 2.0 algorithm in solving SLAM problem.

Recommended Citation

Adel Attia, Mohamed Arafa, Mohamed TALAAT FAHIM, Hadeer (2023) "Enhancing FastSLAM 2.0 performance using a DE Algorithm with Multi-mutation Strategies," Journal of Engineering Research: Vol. 7: Iss. 3, Article 44.
Available at: https://digitalcommons.aaru.edu.jo/erjeng/vol7/iss3/44