NEGL: Lightweight and Efficient Neighborhood Encoding-Based Global Localization for Unmanned Ground Vehicles

This paper proposes a lightweight and efficient Neighborhood Encoding-based Global Localization (NEGL) approach for unmanned ground vehicles (UGVs). To realize the reliable feature description and overcome the restriction of limited Field of View (FOV), neighborhood encoding (NE) scheme is firstly proposed to describe the feature via constructing the structural relationship among neighborhood features. Following that, the reliability of NE scheme is analyzed through the NE-similarity measurement between the priori feature and the detected feature. In addition, the probability model of NEGL is proposed, which is a novel idea based on the priori feature map and simplified through the hierarchical clustering and distance-triggered multiple hypothesis tracking (DT-MHT). Finally, the correct global pose of the vehicle under ambiguous environments is gradually recovered. Comparative experiments using the publicly available datasets and our self-recorded datasets are conducted, and evaluation results show the superior performance of NEGL on success ratio, efficiency, running time and localization accuracy over the adaptive Monte Carlo localization (AMCL), NE+AMCL and Cartographer. Additionally, the experimental results of different FOVs demonstrate NEGL is independent on the range and the direction of the FOV.