Optimization-based non-cooperative spacecraft pose estimation using stereo cameras during proximity operations.

Area of Science:

• Spacecraft dynamics and control
• Computer vision for space applications
• Robotics and autonomous systems

Background:

• Accurate relative pose estimation is crucial for spacecraft proximity operations.
• Stereo-vision is commonly used but suffers from measurement uncertainty.
• Non-cooperative spacecraft pose estimation presents unique challenges.

Purpose of the Study:

• To develop a robust pose estimation method for non-cooperative spacecraft.
• To improve the precision and reliability of stereo-vision measurements in space.
• To address the outstanding issue of uncertainty in relative pose estimation.

Main Methods:

• Utilized binocular structure and spacecraft geometric properties.
• Employed solar panel corner points as distinctive features for pose estimation.
• Implemented stereo matching followed by an optimization-based method for relative pose calculation.

Main Results:

• Achieved a maximum 3D localization error of less than 5%.
• Attained a relative rotation angle error of less than 1 degree.
• Demonstrated improved precision and robustness in both simulations and laboratory experiments.

Conclusions:

• The proposed method offers a significant improvement in spacecraft pose estimation accuracy.
• The use of solar panel geometric constraints enhances the robustness of the system.
• Validated through simulations and laboratory experiments, the method is suitable for real-world space missions.