Research on Factor Graph-based SLAM Localization Algorithm for Multi-source Sensor Fusion

• Published in: 2024 IEEE International Conference on Cognitive Computing and Complex Data (ICCD) pp. 294 - 301
• Main Authors: Ma, Tao, Zhu, Liucun, Wu, Xiao, Chen, Sijie, Wang, Nanxiang
• Format: Conference Proceeding
• Language: English
• Published: IEEE 28.09.2024
• Subjects: Accuracy; Estimation; factor graph; Global navigation satellite system; Laser radar; localization; Location awareness; multi-sensor; Point cloud compression; Robustness; Sensor fusion; Simultaneous localization and mapping; tight coupling; Trajectory
• DOI: 10.1109/ICCD62811.2024.10843557

A multi-sensor tightly coupled localization algorithm based on a factor graph is proposed to address the challenges of low single-sensor localization accuracy and insufficient robustness of mobile robots in outdoor environments. The algorithm incorporates Inertial Measurement Unit (IMU) data at the front end for point cloud de-distortion. It utilizes the IMU pre-integration result as the initial position to enhance point cloud alignment accuracy, thereby improving the overall position estimation of the robot. The back end constructs the IMU pre-integration factor, Lidar odometry factor, Global Navigation Satellite System (GNSS) factor, and loop closure detection factor through a factor graph, and outputs the robot's state information through incremental optimization. Test results on the M2DGR dataset demonstrate that the algorithm significantly enhances localization accuracy and robustness in both closed-loop and open-loop outdoor scenarios.