Extrinsic Parameter Calibration for Camera and Optical Phased Array LiDAR

In autonomous driving, the fusion of camera and LiDAR data is critical for accurate environmental perception, with high-precision extrinsic calibration playing a pivotal role. Optical phased array (OPA) LiDAR, due to its advantages in solid-state scanning, coherent detection, immunity to mechanical fatigue and external interference, and eye safety, represents a promising direction in next-generation LiDAR technology. Conventional LiDAR-camera calibration approaches generally rely on spatial or reflectivity-based point cloud features to infer shared correspondences, followed by nonlinear optimization. However, three key challenges remain: (1) the absence of publicly available datasets for the emerging OPA LiDAR; (2) inaccuracies from sparse point clouds, foreground inflation, and bleeding points affecting feature correspondence; and (3) reliance on complex calibration targets and computationally intensive processes, reducing robustness and efficiency. To overcome these limitations, we propose, for the first time, four joint calibration methods specifically designed for OPA LiDAR. These methods utilize OPA’s directional scanning to treat each scan point as a reliable 3D feature that can be directly matched to corresponding 2D image features, enabling efficient global nonlinear optimization. Experimental validation demonstrates that our methods achieve higher calibration accuracy and significantly reduced computational time compared to existing state-of-the-art techniques. This offers a robust and efficient solution for future multi-sensor fusion systems centered around OPA LiDAR.