This is the official implementation of OmniVIL-Calib, a target-free, mapping-free framework for joint spatiotemporal calibration of an omnidirectional VIL system.
- [2026-01] Our paper has been accepted by IEEE Robotics and Automation Letters (RA-L)!
- [Ongoing] We are currently cleaning up the code and organizing the OmniVIL dataset. Stay tuned!
- Target-free: Calibrate without checkerboards or specific targets.
- Mapping-free: No prior 3D maps or SfM subgraphs required.
- Continuous-time Optimization: Based on B-spline for asynchronous sensor fusion.
- Omni-specific: Specialized motion-flow model for FOV > 180°.
(a) The proposed joint spatiotemporal calibration framework.
(b) The experimental hardware platform and OmniVIL dataset scenarios.
The self-collected dataset used in our paper will be released soon. It includes:
- Synchronized raw data from an omnidirectional camera, IMU, and LiDAR.
- Various indoor and outdoor complex scenarios.
-
Full Dataset Release (ETA: 2026/04/15) Due to hardware limitations at the time of paper submission, hardware-level time synchronization for the camera was not available. We are currently recollecting data with new equipment that supports precise hardware synchronization.
-
Full Code Release (ETA: 2026/05/15) The codebase is relatively large. We plan to first release the dataset and then thoroughly test the code before making it publicly available.
If you find this work useful, please cite:
@article{tu2026omnivil,
title={OmniVIL-Calib: Target-Free Joint Calibration for Omnidirectional Camera, IMU, and LiDAR},
author={Tu, YiHan and Mei, Ruidong and Cheng, Hui},
journal={IEEE Robotics and Automation Letters},
year={2026},
publisher={IEEE}
}