A 6-DOF Navigation Method based on Iterative Closest Imaging Point Algorithm.

Area of Science:

• Spacecraft Navigation
• Computer Vision
• Robotics

Background:

• Autonomous spacecraft navigation requires precise pose estimation.
• Existing vision navigation systems (VNS) often rely on computationally intensive feature matching.

Purpose of the Study:

• To propose a novel algorithm, iterative closest imaging point (ICIP), for six degree-of-freedom autonomous spacecraft navigation.
• To develop a VNS method that overcomes limitations of conventional approaches.

Main Methods:

• The ICIP algorithm estimates full-state pose by finding closest point correspondences between observed and predicted images.
• Initialization is performed using a Gaussian mixture model-based Kalman filter for simultaneous 3D-to-2D point correspondence and camera pose solving.
• The method iteratively optimizes the imaging model without requiring feature matching.

Main Results:

• The proposed navigation method demonstrates high accuracy: three-axis Euler angle within 0.19° (1σ) and three-axis position within 1.87 mm (1σ).
• Simulations and experiments confirm the method's effectiveness, robustness, and computational efficiency.
• The ICIP algorithm shows favorable stability and applicability in simple environments.

Conclusions:

• The ICIP algorithm provides an effective solution for spacecraft autonomous navigation and pose estimation.
• The method's lack of feature matching requirement leads to lower computational complexity and enhanced stability.
• This approach offers a promising alternative for vision-based navigation in space applications.