Event

CalibSwitch: A Camera-LiDAR Extrinsic Recovery and Maintenance Framework

1 reports · 1 sources

Haifa Zhang, Yijing Wang, and Zhiqiang Zuo submitted a paper proposing CalibSwitch, a camera-LiDAR extrinsic calibration framework that separates absolute extrinsic recovery and selective state maintenance into two stages. The framework can operate without providing a numerical initial transformation. On the nuScenes dataset, its CAEG method achieves an average translation error of 0.248 meters and a rotation error of 5.884°. The authors state that paired geometric-observational uncertainty tests, two real vehicle platforms, and execution results at 13.5 FPS on a single-camera Jetson demonstrate the system value of selective geometric state maintenance for continuous perception.