UKF-Based Multi-sensor Data Processing Algorithm for SLAM

Simultaneous localization and mapping (SLAM) algorithm is widely used in unmanned and robotic applications. SLAM usually uses sensor data, which are from the LiDAR, the inertial measurement unit (IMU) and the wheeled odometry, to build maps of the surrounding environment. However, the motion distortion of LiDAR data and the error accumulation of wheeled odometry can affect the accuracy of the front-end odometry of the SLAM algorithm, which can lead to serious errors in the maps created. Therefore, in order to solve the problem of poor map construction results of SLAM algorithm due to various types of sensor errors. we propose the traceless Kalman filtering based (UKF-based) multi-sensor data processing algorithm. Firstly, the proposed algorithm uses the UKF to fuse IMU and wheeled odometry data to reduce the cumulative error of wheeled odometry. And then the algorithm uses IMU and corrected wheeled odometry data and combines with the motion states of the robot to solve the motion distortion problem of LiDAR data. Finally, we verified the effectiveness of the proposed algorithm for wheeled odometry data correction in Gazebo simulation environment and real environment experiments, and compared the mapping effectiveness of SLAM by using the original data and the data processed by our algorithm. The experimental results show that the map construction effect of SLAM algorithm is significantly improved when using the sensor data processed by the algorithm. Experimental validation on the robot shows that the proposed algorithm has universal applicability and good compatibility.