Spacecraft pose estimation gains trustworthy uncertainty from domain-aware calibration
Domain-Aware Conformal Prediction for Trustworthy Spacecraft Pose Estimation under Sim-to-Real Shift
Computer Vision and Pattern Recognition
Summary
Estimating the position and orientation of a spacecraft is important for autonomous space operations, but knowing when those estimates can be trusted is equally vital. The authors found that using a method called conformal prediction calibrated only on simulated images does not work well when applied to real images from space, often missing the true pose silently. They improved this by translating simulated images to look like real ones, which lets the method keep its promises without needing real labels. This approach significantly reduced false confidence and improved reliability for space missions.
What this means in practice
- •For spacecraft navigation teams: Provide trustworthy pose uncertainty estimates during autonomous docking using domain-aware calibration without requiring labeled real images.
- •For industrial robotics engineers: Improve reliability of robot pose estimation under simulation-to-reality shifts by calibrating prediction uncertainties on visually translated simulation data.
Tested on one dataset.
Authors
Nidhal Eddine Chenni, Arunkumar Rathinam, Djamila Aouada
Abstract
Autonomous rendezvous and proximity operations require not only an accurate estimate of the pose of the target spacecraft, but also a reliable indication of when this estimate can be trusted. Conformal prediction provides such an indication with a formal guarantee, by producing pose uncertainty sets that contain the true pose with a specified probability. Such a guarantee, however, assumes that calibration and test data follow the same distribution, an assumption that is violated in orbital perception, where labelled calibration images are available only from synthetic renderings while operational images are real. In this work, we show on the SPEED+ dataset that conformal pose estimation calibrated on synthetic images loses its coverage guarantee on real images, and that many of the resulting failures are silent: the predicted pose set appears valid but excludes the true pose. To address this, we calibrate on synthetic images translated to the appearance of real images. Since translation alters appearance while preserving geometry, the synthetic pose labels remain valid and no labelled real images are required. Experiments with two translation models show that this calibration restores coverage above the specified level for miscoverage levels of 0.1 and above, and reduces silent failures from up to 43.5% to at most 3.3% of the test images, bringing conformal pose estimation a step closer to trustworthy deployment in space.