The Worldwide Journal of Robotics Analysis, Forward of Print.
We suggest AstroSLAM, a standalone vision-based answer for autonomous on-line navigation round an unknown celestial goal small physique. AstroSLAM is based on the formulation of the SLAM drawback as an incrementally rising issue graph, facilitated by way of the GTSAM library and the iSAM2 engine. By combining sensor fusion with orbital movement priors, we obtain improved efficiency over a baseline SLAM answer and outperform state-of-the-art strategies predicated on pre-integrated inertial measurement unit elements. We incorporate orbital movement constraints into the issue graph by devising a novel relative dynamics—RelDyn—issue, which hyperlinks the relative pose of the spacecraft to the issue of predicting trajectories stemming from the movement of the spacecraft within the neighborhood of the small physique. We display AstroSLAM’s efficiency and evaluate towards the state-of-the-art strategies utilizing each actual legacy mission imagery and trajectory knowledge courtesy of NASA’s Planetary Information System, in addition to actual in-lab imagery knowledge produced on a 3 degree-of-freedom spacecraft simulator test-bed.
发布者：Mehregan Dor，转转请注明出处：https://robotalks.cn/astroslam-autonomous-monocular-navigation-in-the-vicinity-of-a-celestial-small-body-theory-and-experiments/
