DreamSat-Pose: Spacecraft Pose Estimation from Single-View 3D Reconstructions and Learned 2D-3D Feature Matching

2026-07-15

AI Summary

This work addresses the joint task of three-dimensional shape reconstruction and six-degree-of-freedom pose estimation for spacecraft from a single image in scenarios with unknown targets. The authors propose an end-to-end framework that first leverages a frozen DINOv2 vision transformer to extract image features and combines them with a trainable dynamic graph convolutional network to reconstruct a 3D point cloud. A dual-stream transformer matcher then performs dense 2D–3D feature alignment, which is fused with a PnP solver to recover high-precision pose estimates. This approach is the first to synergistically integrate a frozen vision transformer with a dynamic geometric network for space object perception, achieving a mean pointing error of 0.157° on the SPE3R dataset using only a single view and demonstrating significantly improved generalization to unseen spacecraft.

Abstract

6-DoF pose estimation is a critical task in autonomous rendezvous and proximity operations. In the case of an unknown target, this task becomes challenging as it shall be paired with the reconstruction of the target shape model. In this article, we propose a novel framework for single-shot shape and pose estimation of unknown spacecraft objects. Given a single image, we first reconstruct a 3D shape model of the target, then estimate the relative six-degrees-of-freedom pose by learning dense 2D-3D correspondences. The image features are extracted using a frozen DINOv3 vision transformer, while the geometric features are computed from the reconstructed point cloud using a trainable dynamic graph convolutional neural network encoder. A dual-stream transformer matcher refines descriptors through alternating self- and cross-attention, producing soft correspondences that are passed to a Perspective-$n$-Point solver for pose recovery. We evaluate the method on the SPE3R dataset and consider FoundationPose as a representative baseline for current state-of-the-art capabilities. Results show reliable pose estimates achieving 0.157 degrees mean pointing error using only a single image and reconstructed geometry, demonstrating strong generalization to unseen spacecraft.

Problem

• spacecraft pose estimation
• 6-DoF pose
• single-view reconstruction
• unknown target
• autonomous rendezvous

Innovation

• 6-DoF pose estimation
• single-view 3D reconstruction
• 2D-3D feature matching
• vision transformer
• dynamic graph CNN