Search papers, labs, and topics across Lattice.
This paper introduces a novel method for extrinsic calibration of non-overlapping multi-camera systems that utilizes pure rotational motion and a single static calibration board, circumventing the limitations of conventional target-based and motion-based methods. By employing a global optimization framework with a latent turntable frame, the authors effectively integrate time-separated observations from multiple cameras observing the same target. Extensive experiments demonstrate that this approach achieves competitive accuracy and robustness, making it suitable for practical deployments without the need for specialized hardware.
Achieving accurate multi-camera calibration without overlapping fields of view or specialized hardware could revolutionize camera setups in robotics.
Extrinsic calibration of multi-camera systems with non-overlapping FOVs has been a challenging problem in the robotics literature. Conventional target-based methods impose substantial target setup overhead, either deploying large calibration targets or requiring pre-measured multi-target poses. Motion-based approaches instead suffer from drift error, scale ambiguity, and motion degeneracy. Securing both accuracy and usability, we propose a novel calibration method that leverages pure rotational motion, requiring only a single static calibration board. The key idea is to make all cameras sequentially observe the same target under a shared geometric reference, even without overlapping views. To integrate these time-separated observations, we formulate the problem using a latent turntable frame and a 3D error on SE(3) within a global optimization framework. We validate the proposed method on both a controlled camera rig and a full-scale vehicle platform with heterogeneous cameras, and analyze robustness under non-ideal turntable motion. Extensive experiments show that our approach maintains competitive accuracy without specialized precision hardware, proving its strong suitability for realistic on-site deployments. Our code is publicly available here.
