Monocular Pose Estimation of an Uncooperative Spacecraft Using Convexity Defect Features.

Area of Science:

• Robotics
• Computer Vision
• Aerospace Engineering

Background:

• Spacecraft relative pose estimation is crucial for autonomous proximity operations.
• Uncooperative targets lack sensors or markers, complicating pose estimation.
• Initial pose estimates are often unavailable, posing a significant challenge.

Purpose of the Study:

• To develop a novel monocular pose estimation algorithm for uncooperative spacecraft.
• To address challenges in pose initialization for known, uncooperative targets.
• To enable accurate relative navigation for applications like on-orbit servicing.

Main Methods:

• Utilizes convexity defect features from target images for feature matching.
• Matches image features to points on a known target spacecraft model.
• Estimates pose by solving the Perspective-n-Point (PnP) problem.

Main Results:

• Simulations demonstrate accurate pose estimation across varying relative attitudes and distances.
• Algorithm performance is validated with changes in solar panel dimensions.
• Robust feature detection is confirmed as critical for accurate pose estimation.

Conclusions:

• The proposed algorithm effectively addresses pose initialization for uncooperative spacecraft.
• It provides a viable solution for relative navigation in proximity operations.
• This method supports critical applications such as on-orbit servicing.