Point Cloud Distortion Correction in Mobile Robots Using Inertial Measurement Unit Integration

This paper proposes a method to correct point cloud distortions caused by rotating LiDAR used in robots such as drones, utilizing an Inertial Measurement Unit (IMU). LiDAR works by emitting light, measuring the time it takes for the light to reflect off objects and return, and then calculating the distance. When a rotating LiDAR is mounted on a robot, the input point cloud data is significantly distorted during the robot's rotation compared to solid-state LiDAR. To correct this distortion, our study leverages the time information of each point's measurement, proposing a method to adjust the distorted point cloud data based on the start time of the LiDAR scan. By utilizing IMU sensor data, we track the dynamic movements of the robot and use the measured time information to correct the distortions caused by the robot's rotation. The proposed method was validated using the NTU VIRAL dataset in conjunction with FAST-LIVO. Experimental results show that applying our method reduces the Absolute Trajectory Error (ATE) by an average of 0.18 meters, with a maximum reduction of up to 0.49 meters. This improvement is expected to contribute to the enhancement of LiDAR data accuracy and the advancement of technologies for precise robot localization and environmental perception.