Vision-based sampling implementation in the Chang’e-6 lunar farside sample return mission

Lunar sample return missions are crucial for researching the composition and origin of the Moon. In recent decades, several lunar sample return missions have been conducted, yielding abundant and valuable lunar samples. As the latest development in lunar sample returns, the Chang’e-6 mission aimed to implement lunar farside sampling. The shorter time available for sampling requires higher sampling efficiency. In this study, the main factors in the sampling site selection and sampling process are introduced and a vision-based sampling implementation is designed for the Chang’e-6 mission to significantly simplify manual operation while maintaining high sampling quality. By sufficiently leveraging the point cloud data reconstructed from the binocular camera images, autonomous terrain analysis and sample point selection are achieved. A 6D pose estimation pipeline based on point cloud registration provides a robust method for sampler pose measurement, replacing the previous manual fine-tuning process and achieving better accuracy. Owing to the well-analyzed sample points and accurate fine-tuning, the proposed approach demonstrates high accuracy in controlling the scooping depth, while significantly reducing the time cost of the sampling implementation, effectively supporting the Chang’e-6 lunar sample mission.