Rendezvous and docking remains a demanding space operation. Historically, performed with strong ground supervision and monitoring in Earth or lunar orbit, these missions also tend to be "point solutions"; specific to the target vehicle and orbit. This approach manifests high cost for mission preparation and mission operations, and long lead for development for new rendezvous missions. In turn, these issues have made debris removal missions prohibitively expensive and have created the inability to rapidly respond to new debris threats or changing environments. For this Phase I, Atomos, together with The Charles Stark Draper Laboratory, Inc. (“Draper”) will accelerate the development of systems that allow automated rendezvous with diverse targets, including a machine-learning based pose and position estimator, and on-board rendezvous approach and abort planner. Atomos will leverage previous developments (planned to be demonstrated on-orbit in 2023), including a GPU-based image landmark processing and recognition program developed for tracking fiducials, extended to debris features; a Kalman Filter designed to fuse the data from multiple image and ranging streams; and a rendezvous trajectory targeting system, as the starting point for this effort.
