Kongzhi Yu Xinxi Jishu (Feb 2024)
A Method of Simultaneous Vehicle Localization and Roadside Pole-shaped Object Inventory Creation Based on Integrated LiDAR-IMU-GNSS System
Abstract
At present, the use of mobile LiDAR systems (MLS) to collect environmental information and generate roadside pole-shaped object inventories is limited by equipment costs and has poor real-time performance. Although LiDAR-based simultaneous localization and mapping (SLAM) techniques have been widely applied in the navigation field, research on simultaneous real-time localization and creation of roadside pole-shaped object inventories remains absent. In this regard, this paper proposes an approach that utilizes LiDAR technology to achieve vehicle localization and create roadside pole-shaped object inventories complete with absolute locations. The aim is to devise an accurate and robust system for vehicle localization and the creation of roadside pole-shaped object inventories. Firstly, LiDAR was integrated with inertial measurement units (IMU) and global navigation satellite system (GNSS), to achieve accurate pose estimations and simultaneous generation of global maps. Secondly, an optimized fusion positioning algorithm based on sliding windows is constructed, to enhance system robustness through effectively integrating data from multiple sensors. Then, a method for creating pole-shaped object inventories is proposed using an SLAM feature extraction algorithm, thereby reducing the computational expense for simultaneous vehicle positioning and roadside pole-shaped object inventory creation. Finally, extensive evaluations are conducted using real datasets covering various road scenarios, including urban and suburban areas. The experimental results demonstrated the centimeter-level vehicle positioning accuracy of the proposed system in real-time automatic creation of roadside pole-shaped object inventories, boasting an average positioning error within 3 cm.
Keywords