LiDAR-Based Non-Cooperative Tumbling Spacecraft Pose Tracking by Fusing Depth Maps and Point Clouds

Area of Science:

• Robotics and Autonomous Systems
• Aerospace Engineering
• Computer Vision

Background:

• Space proximity operations require precise relative pose determination between spacecraft.
• Tumbling target spacecraft present significant challenges for real-time pose estimation.
• Existing methods struggle with high-speed tumbling and close-range scenarios.

Purpose of the Study:

• To develop a robust LiDAR-based method for real-time relative pose tracking between a chaser and a tumbling target spacecraft.
• To address the challenges of close-range operations and significant pose variations.

Main Methods:

• Fusing LiDAR-derived depth maps and point clouds for pose estimation.
• Utilizing line detection and matching in depth maps to estimate roll angle variations.
• Employing adaptive voxelized grid simplification for real-time point cloud processing.
• Applying the Iterative Closest Point (ICP) algorithm for point cloud registration.

Main Results:

• Demonstrated capability in estimating real-time 6-DOF (Degrees of Freedom) relative pose.
• Successfully handled large pose variations and simulated tumbling motions.
• Validated through numerical experiments simulating space proximity missions.

Conclusions:

• The proposed LiDAR-based method offers a viable solution for challenging spacecraft relative pose determination.
• The fusion of depth maps and point clouds enables accurate tracking of high-speed tumbling targets.
• The method satisfies real-time requirements for space proximity and capture missions.