Temporal and Spatial Online Integrated Calibration for Camera and LiDAR

TL;DR

This paper presents an online calibration method for camera and LiDAR sensors that enhances data fusion accuracy and real-time performance through pose estimation, environmental robustness, and dynamic target elimination, validated on KITTI benchmark.

Contribution

It introduces a novel online calibration approach combining pose estimation and robust feature extraction for real-time camera-LiDAR synchronization and spatial calibration.

Findings

Improves temporal calibration accuracy by 38.5% over soft synchronization.

Automatically corrects disturbance errors within 0.4 seconds.

Achieves spatial calibration accuracy of 0.3 degrees.

Abstract

While camera and LiDAR are widely used in most of the assisted and autonomous driving systems, only a few works have been proposed to associate the temporal synchronization and extrinsic calibration for camera and LiDAR which are dedicated to online sensors data fusion. The temporal and spatial calibration technologies are facing the challenges of lack of relevance and real-time. In this paper, we introduce the pose estimation model and environmental robust line features extraction to improve the relevance of data fusion and instant online ability of correction. Dynamic targets eliminating aims to seek optimal policy considering the correspondence of point cloud matching between adjacent moments. The searching optimization process aims to provide accurate parameters with both computation accuracy and efficiency. To demonstrate the benefits of this method, we evaluate it on the KITTI benchmark with ground truth value. In online experiments, our approach improves the accuracy by 38.5\% than the soft synchronization method in temporal calibration. While in spatial calibration, our approach automatically corrects disturbance errors within 0.4 second and achieves an accuracy of 0.3-degree. This work can promote the research and application of sensor fusion.