Realtime and Integrated Framework for LiDAR-based Object Tracking

This study proposes a real-time integrated framework for LiDAR-based object tracking in autonomous driving environments. Advancements in LiDAR sensors are increasing point cloud data collection, leading to a demand for reliable real-time processing methods. The proposed framework applies voxelization and ground removal techniques to reduce computational load and integrates clustering and deep learning-based object recognition to ensure stability. Combining the point cloud data from LiDAR and the IMU data corrects distortions and refines real-time object movement, enabling accurate tracking in dynamic environments. This framework supports a maximum detection range of 100 m, with a computation time of 52 ms, a positional error of 1.06 m, a heading error of 3.79°, a relative velocity error of 1.46 m/s, and an average tracking frame count of 101, thereby improving object recognition accuracy and tracking performance while fulfilling real-time processing requirements.