A Robust Lidar SLAM System Based on Multi-Sensor Fusion

In this paper, we propose a LiDAR-based multi-sensor fusion SLAM system that integrates magnetometer, odometer and IMU information to solve the problem of accuracy degradation of lidar SLAM algorithm in scenes with insufficient structural features. In the lidar odometer part, based on the feature-based point cloud matching algorithm, magnetometer and odometer constraints are introduced to improve the robustness of the algorithm. At the back end, we constructed a factor graph for the global pose optimization, and added the measurement information of each sensor into the factor graph as a factor, so as to realize the nonlinear optimization of the pose and IMU bias. Experimental results show that the proposed algorithm has good robustness and accuracy, and is superior to LeGO-LOAM algorithm in positioning error.

I. Introduction

Simultaneous localization and map building (SLAM) is one of the key research contents in the field of mobile robots, and it is the key technology to realize autonomous navigation of robots in unknown environments [1]. According to the types of sensors used, SLAM algorithms can be divided into lidar SLAM and visual-based SLAM [2]. Visual SLAM is easily affected by light and texture, which leads to the decrease of accuracy and the lack of all-weather working ability [3]. Lidar SLAM, which takes lidar as the main sensor, is not limited by illumination, and has high accuracy and stability [4], so it has been widely used in the field of navigation and positioning. However, lidar SLAM algorithms is prone to precision degradation when working in scenes with insufficient geometric features [5].