Angles-Only Relative Navigation Activities during AVANTI

Kurzfassung

This contribution addresses the realization of angles-only relative navigation systems as means to approach noncooperative target objects flying in low Earth orbits. Based on the recent in-flight experience collected during the AVANTI (Autonomous Vision Approach Navigation and Target Identification) demonstration, a critical comparison between spaceborne and ground-based design philosophies is drawn. Focus is given on the motivations behind the choice of the employed techniques, as well as on the consequent attainable performances. A pure vision-based angles-only approach, in fact, represents an appealing strategy for future on-orbit servicing and debris removal missions, since it requires simply a passive camera as sensing instrument. However, this comes at the cost of a weakly observable relative orbit determination problem, which demands special care, especially during autonomous operations. For the first time in space applications, AVANTI demonstrated the capability to autonomously navigate towards a fully noncooperative target satellite in low Earth orbit making use of angles-only measurements from 50km to circa 50m of inter-satellite separation range. Within AVANTI, the DLR Earth-observation BIROS spacecraft performed far- to mid-range proximity operations with respect to the BEESAT-4 one-unit CubeSat (Berlin Technical University), released in orbit on the 9th of September 2016, by means of a single picosatellite launcher device. To meet these goals, AVANTI employed the star-tracker embarked on BIROS as far-range camera to take images of portions of the sky and autonomously carried out onboard the following activities: image processing and target identification to provide the angles-measurements of the line-of-sight to the target; real-time relative navigation using an extended Kalman filter and computation of the required impulsive maneuvers' profile to perform a rendezvous in a safe, fuel efficient manner. The peculiarity of the AVANTI demonstration compared to the formation-flying missions flown so far in low Earth orbit is that it was confronted to an incontrovertible noncooperative scenario. Although BEESAT-4 embarks a Phoenix GPS receiver, in fact, such device was not yet commissioned and, therefore, not operating by the time when AVANTI took place. Radar-tracking observations could not be used when the satellites were separated by less than 5 km, due to the impossibility to distinguish the signals emitted from the two spacecraft. Two-line elements, which generally are not accurate enough to support close-range proximity operations, are also affected by the same problem at close range. Thus, the only observations available during AVANTI were the pictures collected by BIROS's star-tracker, and consequently, the only way to monitor and cross-evaluate the behavior of the spaceborne solution was to re-process such measurements a-posteriori on-ground. Contrary to the onboard unit, the ground-based navigation system could benefit from the availability of calibrated maneuvers from GPS-based precise absolute orbit determination, and, of course, from the presence of man in-the-loop. In addition, alternative image processing and filter techniques (i.e., iterative and/or batch schemes) could be employed, given the lack of real-time and computational resources constraints. Regarding both spaceborne and ground-based navigation systems, AVANTI capitalizes the experience already collected in 2012 using the PRISMA formation flying testbed. At that time, the so-called ARGON (Advanced Rendezvous demonstration using GPS and Optical Navigation) experiment had already tackled the problem of angle-only relative navigation by making a ground-in-the-loop approach to a target using optical methods. With respect to such achievements, AVANTI demanded an increased level of complexity to cope with a more challenging mission scenario. Contrary to ARGON which, thanks to the dusk-dawn orbit of PRISMA, benefited from optimal illumination conditions, AVANTI experienced frequent and extended interruptions of visual tracking. Such irregular visibility conditions derive by the fact that the target and chaser spacecraft are eclipsed by Earth during a large part of their orbit and on the other hand the camera becomes blinded by the Sun during another large part of the orbit. In addition, BIROS flies at a low altitude (500 km) inducing a strong unknown differential drag which has to be estimated as part of the orbit determination. Combined with the fact that the picosatellite is a tiny object and that the problem is weakly observable, these constraints make the angles-only relative orbit determination very challenging. Despite all the aforementioned difficulties, flight data show that the filter design retained for AVANTI was perfectly suited for the needs of the experiment. Two approaches have been performed autonomously: from 13 km to 1 km (Nov 19-23, 2016) and from 3 km to 30 m (Nov 25-28, 2016). Initialized from the ground with a reasonably good guess of the relative state, the filter was able to support the onboard controller with a navigation solution accurate at the meter level in the lateral direction and to about 10% of the inter-satellite separation in the boresight direction. Considering different stages of experiment commissioning, and phases with rising levels of autonomy, almost two months of flight data support the proposed analysis. Indeed AVANTI demonstrated the viability of the angles-only navigation approach in a general, and thus extremely representative, orbit scenario. Nevertheless, the Authors shed light on critical points still deserving improvements and propose a critical assessment of the performances achievable in flight.