An evidential theory is widely used for 2D grid-based localization in a robotics field because the theory has benefits to consider additional states such as 'unknown' and 'conflict'. However, there are some problems such as computational limitation and excessive resource share when the localization system is expanded from 2D grid to 3D voxel. In order to overcome the problems, this paper proposes the parallelized particle filter based localization system using 3D evidential voxel maps. A many-core processor based parallel computing framework with optimization techniques is applied to accelerate the computing power. Experiments were performed to evaluate the performance of the localization system in a complex environment, and to compare the computational time and resources between various types of processing units. The experimental results show that the proposed parallel particle filter is much more efficient than particle filter without parallel computing regarding computational cost.
A GPU Accelerated Particle Filter Based Localization Using 3D Evidential Voxel Maps
Sae Technical Papers
WCX SAE World Congress Experience ; 2019
2019-04-02
Conference paper
English
Monocular localization within sparse voxel maps
IEEE | 2017
|G-VOM: A GPU Accelerated Voxel Off-Road Mapping System
IEEE | 2022
|Multi-sensor environment perception for automated vehicles with semantic evidential grid maps
TIBKAT | 2023
|