¹Research Centre of Electromechanical Technology, Zhejiang Normal University, Jinhua, Zhejiang, China
²College of Computer Science and Technology, Nanjing University of Science and Technology, Nanjing, China

---

Abstract—The joint state is factored into a path component and a map component in the FastSLAM framework, which reduces the computational complexity greatly. Among all the filtering solutions based on the FastSLAM framework, the particle filtering solution and the kalman filtering solution are the two most important series. Analysis of the two filtering SLAM solutions is made in this paper. And the conclusions are as follows: Firstly, the kalman filtering SLAM solution is superior to the particle filtering SLAM solution in computational complexity. Secondly, the two filtering SLAM solutions perform similarly well in estimation accuracy. In addition, the kalman filtering SLAM solution and the particle filtering SLAM solution are evaluated with the ‘Victoria Park Dataset’, which is a bench mark dataset in the SLAM community. The experimental results show that the association between observations and estimated features is vital to the accuracy and convergence of the FastSLAM framework based solutions. Thus, special concern should be paid to reliability and stability of the data association process in SLAM.

Index Terms—Simultaneous Localization and Map Building (SLAM), FastSLAM; Particle Filter, Kalman Filter, Analysis.

[PDF]

Cite: Wu Zhou, Chunxia Zhao, "Analysis of filtering solutions based on the ‘FastSLAM’ framework," Journal of Computers vol. 7, no. 5, pp. 1169-1175, 2012.