Research on Laser SLAM Algorithm for Multi sensor Fusion Based on Elastic Tight Hybrid Coupling

In response to complex environments such as underground caves, pipelines, and tunnels without lighting and GPS, and the low accuracy and robustness of robot positioning and mapping caused by structural degradation, motion turbulence, and sensor degradation, this paper proposes a factor graph optimized pose estimation and mapping framework based on the loose tight hybrid coupling of lidar inertial navigation wheel encoder. This framework uses IMU to estimate motion state and correct laser point cloud distortion. The front-end extracts and matches features based on line and surface features. In order to reduce computational complexity, keyframe extraction and subgraph construction methods are used; The backend is optimized by constructing laser odometer factors, IMU predicted sub factors, wheel encoder factors, and loop detection factors based on a factor graph model. We evaluated the performance of the system through an open source public dataset, and the test results showed that the absolute trajectory error of the LIW-SLAM algorithm proposed in this paper is 1.278m, with an error percentage of 0.932%, which is better than the 2.206% of LIO-SAM. The results indicate that the fusion framework of this system has high positioning accuracy, high robustness, and good generalization ability in the case of turbulent motion and structural degradation.