Point Cloud Based Relative Pose Estimation of a Satellite in Close Range

Area of Science:

• Robotics and Space Engineering
• Computer Vision and Sensor Fusion

Background:

• Relative pose estimation is critical for autonomous space operations like rendezvous and servicing.
• Existing methods often struggle with sensor limitations and computational efficiency in close-range scenarios.

Purpose of the Study:

• To develop and validate a novel model-based pose estimation method for target satellites using flash LIDAR point cloud data.
• To address challenges in pose initial acquisition and tracking for close-range satellite operations.

Main Methods:

• A novel pose estimation approach combining global optimal searching for initial acquisition and Iterative Closest Point (ICP) for tracking.
• Utilizing dense point cloud data directly from a flash LIDAR sensor.
• Development of a simulation system for sensor performance evaluation and ground truth generation.

Main Results:

• The proposed method demonstrates robust performance with direct point cloud processing and accommodates large pose variations.
• Numerical simulations confirmed the algorithm's effectiveness and quantifiable pose accuracy.
• Field testing validated the practical applicability and effectiveness of the developed pose estimation technique.

Conclusions:

• The novel model-based pose estimation method is effective for determining the relative pose of satellites in close proximity.
• The approach shows promise for enhancing the autonomy and safety of space rendezvous and servicing missions.
• Direct processing of flash LIDAR point clouds offers a viable solution for real-time pose estimation in space applications.