Technology Category
Computer Vision: Diffusion Models for VisionMachine Learning: Calibration & Uncertainty QuantificationSearch and Optimization: Learning to Search
Iterative Camera-LiDAR Extrinsic Optimization via Surrogate Diffusion
🤖 AI Summary
To address the low accuracy and lack of iterative optimization capability in camera–LiDAR extrinsic calibration for autonomous driving, this paper proposes the first end-to-end, single-model iterative optimization framework based on surrogate diffusion. We innovatively design a surrogate diffusion mechanism coupled with a buffered inference strategy, and construct a dual-path (projection-first/encoding-first) denoising network to enhance point cloud feature modeling—enabling multi-scale calibration accuracy within a unified model. Experiments demonstrate that our method reduces rotational error by 24.5% over the second-best approach; compared to a baseline diffusion scheme, it achieves 20.4% and 9.6% reductions in rotational and translational errors, respectively, while reducing inference latency by 43.7%. This work pioneers the integration of diffusion models into iterative extrinsic calibration, uniquely balancing calibration accuracy, computational efficiency, and deployment feasibility.
Technology Category
Application Category
📝 Abstract
Cameras and LiDAR are essential sensors for autonomous vehicles. Camera-LiDAR data fusion compensate for deficiencies of stand-alone sensors but relies on precise extrinsic calibration. Many learning-based calibration methods predict extrinsic parameters in a single step. Driven by the growing demand for higher accuracy, a few approaches utilize multi-range models or integrate multiple methods to improve extrinsic parameter predictions, but these strategies incur extended training times and require additional storage for separate models. To address these issues, we propose a single-model iterative approach based on surrogate diffusion to significantly enhance the capacity of individual calibration methods. By applying a buffering technique proposed by us, the inference time of our surrogate diffusion is 43.7% less than that of multi-range models. Additionally, we create a calibration network as our denoiser, featuring both projection-first and encoding-first branches for effective point feature extraction. Extensive experiments demonstrate that our diffusion model outperforms other single-model iterative methods and delivers competitive results compared to multi-range models. Our denoiser exceeds state-of-the-art calibration methods, reducing the rotation error by 24.5% compared to the second-best method. Furthermore, with the proposed diffusion applied, it achieves 20.4% less rotation error and 9.6% less translation error.
Problem
Research questions and friction points this paper is trying to address.
Improves camera-LiDAR extrinsic calibration accuracy
Enables iterative optimization via surrogate diffusion
Enhances existing calibration methods without architectural changes
Innovation
Methods, ideas, or system contributions that make the work stand out.
Iterative extrinsic refinement via surrogate diffusion
Enhances calibration methods without architectural changes
Uses denoising process for higher accuracy robustness
🔎 Similar Papers
2024-01-31arXiv.orgCitations: 6
