In context
By 2023, multi-sensor fusion had become essential for autonomous driving and mobile robotics, with LiDAR and IMU tightly coupled in SLAM systems. However, accurate calibration of the external parameters between LiDAR and IMU remained a bottleneck, especially in environments with occlusions or lacking large planar surfaces, where existing methods like LI-Calib struggled.
What was reported
Researchers at Zhejiang University proposed a robust LiDAR-IMU joint calibration method that improves accuracy in complex environments. The method introduces line features, which are less affected by occlusions and provide precise localization, alongside planar patch features to strengthen constraints during matching construction.
A two-stage iterative optimization pipeline is employed, where line feature constraints are added only after initial convergence. Adaptive loss weights, based on geometric residuals from each iteration, help avoid local minima and improve convergence.
Tests on both an open-source outdoor dataset and a self-built indoor dataset showed a calibration standard deviation of about 2 mm for translation and 0.04 degrees for rotation, outperforming current state-of-the-art methods.
Why it mattered
This work addressed a practical gap in LiDAR-IMU calibration, enabling reliable sensor fusion in real-world environments such as cluttered indoor spaces and urban roads. The improved accuracy and robustness support higher-performance SLAM and perception systems for autonomous vehicles and robots, advancing the deployment of multi-sensor platforms in industrial automation.
“The proposed method is tested with the open-source outdoor dataset and the self-built indoor dataset. The results show that the calibration standard deviation of the proposed method for translation external parameters is about 2 mm and the calibration standard deviation for rotating external parameters is about 0.04 degrees.”
Source: 《机器人》期刊 (robot.sia.cn) · Published 2023-05-09 · “一种鲁棒的LiDAR-IMU联合标定方法”