ACSC: Automatic Calibration for Non-repetitive Scanning Solid-State LiDAR and Camera Systems (2011.08516v1)

Abstract: Recently, the rapid development of Solid-State LiDAR (SSL) enables low-cost and efficient obtainment of 3D point clouds from the environment, which has inspired a large quantity of studies and applications. However, the non-uniformity of its scanning pattern, and the inconsistency of the ranging error distribution bring challenges to its calibration task. In this paper, we proposed a fully automatic calibration method for the non-repetitive scanning SSL and camera systems. First, a temporal-spatial-based geometric feature refinement method is presented, to extract effective features from SSL point clouds; then, the 3D corners of the calibration target (a printed checkerboard) are estimated with the reflectance distribution of points. Based on the above, a target-based extrinsic calibration method is finally proposed. We evaluate the proposed method on different types of LiDAR and camera sensor combinations in real conditions, and achieve accuracy and robustness calibration results. The code is available at https://github.com/HViktorTsoi/ACSC.git .

• The paper presents a fully automated calibration technique that refines sparse SSL point clouds using temporal-spatial integration and feature enhancement.
• It employs 3D corner estimation paired with a PnP strategy to accurately align LiDAR and camera data, minimizing calibration errors.
• Comparative evaluations show the method achieves sub-pixel reprojection errors, outperforming prior approaches in diverse sensor setups.

Overview of ACSC: Automatic Calibration for Non-repetitive Scanning Solid-State LiDAR and Camera Systems

The paper introduces an innovative approach to sensor calibration between Solid-State LiDARs (SSL) and camera systems, focusing on addressing the non-uniformities in SSL scanning and ranging errors that complicate the calibration task. The authors propose the ACSC method, a fully automated process designed to enhance the efficiency and accuracy of calibrating these systems by combining SSL and camera data through a novel method that makes use of time-domain integration and feature refinement.

Methodology

The core contribution of the paper is the automatic, target-based calibration technique for SSL-camera systems. The method entails three primary steps:

1. Feature Refinement: The SSL's non-repetitive scanning patterns create sparse and noisily distributed point clouds that require enhancement. The paper describes a method for temporal-spatial geometric feature refinement by integrating point clouds over time to densify them while applying noise reduction techniques. This is followed by a projective transformation of the better fit the points onto a hypothetical ideal plane that represents the calibration target.
2. 3D Corner Estimation: Utilizing the unique reflectance intensity pattern of the SSL-observed calibration target, this method optimizes the alignment of observed data to a predefined standard model using a non-linear optimization method. The reflectance pattern, which closely follows the black-and-white grid patterns typically found on calibration checkerboards, is key to accurately determining the 3D position of corners in SSL data.
3. Extrinsic Calibration: By correlating 3D corners from LiDAR data with 2D corners identified in camera images, the ACSC uses a PnP-based approach to compute the extrinsic matrix linking the LiDAR and camera coordinate frames. This step is complemented by a method for iteratively solving and validating fitting parameters, discarding outliers and improving set results with each iteration.

Results and Evaluation

The proposed ACSC method is evaluated using various combinations of Livox SSL models and different cameras. Visualization of reprojection errors—projecting the SSL-based 3D point data back into the 2D image plane for comparison with directly derived image data—demonstrated promising accuracy, with most systems achieving normalized reprojection errors well within sub-pixel ranges (typically under 0.6 pixel for several SSL models).

Quantitative evaluation highlights that the ACSC method outperformed prior works such as MI and ILCC, particularly in contexts highly suited to SSL configuration characteristics, where traditional methods often struggled. The ACSC method demonstrated consistent calibration results across different setups, showing robustness to variability in sensor qualities and placements.

Implications and Future Directions

The ACSC method presents a significant step forward in the practical deployment of SSL-camera systems for burgeoning fields like autonomous driving and robotic vision. By simplifying calibration processes and enhancing accuracy with consideration to SSL idiosyncrasies, the method enables better sensor fusion, which is crucial for tasks demanding high precision, such as object detection, mapping, and tracking.

Looking forward, further improvements could be explored by extending this approach to more dynamic calibration scenarios, potentially paving paths toward real-time, on-the-go calibration in fluctuating environments. Moreover, the method could benefit from deeper analysis of SSL characteristics in changing light or weather conditions, enriching capability and robustness beyond controlled settings. Additionally, integrating AI-based pattern recognition might automate and fine-tune the reflective distribution alignment process, reinforcing the bridge between robust theoretical underpinnings and adaptive real-world applications in next-gen autonomous systems.