Online Extrinsic Calibration of Camera and LiDAR Based on Cascade Optimization

Area of Science:

• Robotics
• Computer Vision
• Sensor Fusion

Background:

• Accurate extrinsic calibration is crucial for camera and LiDAR sensor fusion and positioning.
• Traditional targetless methods struggle with scene adaptability and convergence stability in complex environments.

Purpose of the Study:

• To develop an online, cascade-optimization-based extrinsic calibration method for camera-LiDAR systems.
• To overcome limitations of traditional methods, enhancing accuracy and robustness in challenging scenarios.

Main Methods:

• Implemented an initial hand-eye calibration using camera and LiDAR odometry alignment.
• Utilized depth differences and planar edges for robust edge line feature extraction from sparse point clouds.
• Applied Principal Component Analysis (PCA) for decoupled optimization based on directional observability.

Main Results:

• Achieved rotational parameter errors within 0.25° and translational errors below 0.05 m on urban datasets.
• Demonstrated method's generalizability and competitive accuracy.
• Effectively mitigated issues from uneven environmental feature distribution.

Conclusions:

• The proposed method provides accurate and stable extrinsic calibration for camera-LiDAR systems.
• It is suitable for high-accuracy engineering applications requiring robust sensor fusion and positioning.