Box Particle Filtering for SLAM with Bounded Errors

Peng Wang (1) , Philippe Xu (2) , Philippe Bonnifait (2) , Jianwen Jiang

1 Southwestern University of Finance and Economics [Chengdu, China]
2 Heudiasyc - Heuristique et Diagnostic des Systèmes Complexes [Compiègne]

This paper proposes a set-membership based method for Simultaneous Localization and Mapping (SLAM). A Box Particle Filter (BPF) is exploited and improved to estimate robot states and feature positions, with interval Constraint Propagation (CP) to reduce box sizes and decrease uncertainties in estimates. Buffers are also used to get q-satisfied results when empty estimations arise, on the one hand. On the other hand, through buffer contraction, historical estimations can be improved. Illustrations of the proposed method are given over simulations and experiments, with comparisons with a Particle Filter (PF) based method. The results show that the proposed method can reach the same SLAM accuracy as PF based method with much fewer particles. Moreover, this approach is more robust to high level uncertainties.

Domaines: Robotique [cs.RO]

Peng Wang, Philippe Xu, Philippe Bonnifait, Jianwen Jiang. Box Particle Filtering for SLAM with Bounded Errors. 15th International Conference on Control, Automation, Robotics and Vision (ICARCV 2018), Nov 2018, Singapore, Singapore. pp.1032-1038, ⟨10.1109/ICARCV.2018.8581234⟩. ⟨hal-02060133⟩