Joint Multi-Camera LiDAR Extrinsic Calibration via Learned Pairwise Initialization and Geometric Refinement
Researchers have developed a novel two-stage framework for joint multi-camera LiDAR extrinsic calibration. This method first uses CMRNext for initial pairwise estimates and then refines them using a multi-frame bundle adjustment. The approach ensures system-level consistency by accounting for the geometric coupling in multi-camera setups, outperforming independent pairwise methods on both in-domain and out-of-domain datasets. AI
IMPACT Improves sensor fusion accuracy for autonomous systems by ensuring consistent calibration across multiple cameras and LiDAR sensors.