(Duration of 20 days)
The ARV 2 is the second of the two ARV experiments. Here, the initial relative distance between the two satellites is 30 km. The experiment relies solely on the VBS system to find and rendezvous with the Tango satellite.
The experiment lasted almost 30 hours, part of which not possible to monitor from ground when no visibility from the ground segment was available. Below is a video from the initial Target Search manoeuvre. Mango is scanning the sky with the VBS, until it detects Tango (at about 17:47) and it centers it in the camera. Then the process of Orbit Determination starts. At about one minute in the video one can see a false detection of Tango, but the GNC system in Mango ignores it and continues. Next at about 2:20 in the video, Tango is detected and centered.
The Orbit Determination starts with an initial guess of the distance to Tango, which is 6km further away then it actually is. The difficulty for Mango is that the VBS only gives a direction towards Tango, but not the range as the camera can not see that. So with angles only information, Mango now has to estimate the range. In about 3 orbits, the 6km error has been reduced to 2km, and the spacecraft can start aligning the orbits. It takes all decisions autonomously, and does not require a go ahead from ground to fire its thrusters. After the Orbit Alignment is completed, Mango is chasing Tango at a distance of 30km but otherwise in practically the same orbit. As the distance is so large it performs two rather large thrusts, increasing the relative speed to about 5km per orbit. When it gets closer then 15km, the guidance starts to control towards predefined distances. The final orbits decrease the distance with Tango with a factor of 3: 1350m, 450m, 150m and finally 50m. Here Mango stays for an orbit, station keeping, until ground commands it to the AFF mode, ready for another experiment.
Below is a plot showing the Orbit Determination results: the position of Mango relative to Tango in three dimensions (X is range) in red according to GPS, and in blue according to the onboard VBS navigation filter, processing results from the VBS sensor.
Below is a plot showing the actual distance and the successive references (in red) that Mango is attempting to realise.
The plot below shows the final delivery in three axes. X is the distance, which can be seen to be maintained to 50m, with a tolerance of a few meters. Note that no range measurement is involved, so it is challenging to stay at the same spot. Tango is actually quite large on the VBS CCD from this distance and the measurements are therefore of quite low quality. It can be seen how the navigation is kicked around at two locations, where the spacecraft thinks it is a few meters to the side. Nevertheless, the system was robust enough to tolerate this and kept itself in place.
Below is an image that was taken with the VBS Far Range camera from 50m. As can be seen, Tango is quite large, and several features are visible.
The experiment demonstrated:
- autonomy, in that no operator was required, nor time scheduled telecommands after the experiment was set in motion;
- autonomous detection of Tango, and initialization of the navigation filters, with an initial guess for the state that was 6km off in range, 300m in cross track, 200m in radial, and substantial error in the phasing of the orbit;
- autonomous orbit determination of Target, reducing the range error to 2km (best performance seen in earlier ARV was 1km), sufficient to predict the orbit of Tango for the rest of the experiment and navigate with respect to it;
- autonomous aligning of the orbits to within the prescribed tolerances, which also held when checking them with GPS;
- robustness against severe measurement errors in the VBS: this experiment suffered from the tracking of a bright star by the VBS, which was rejected by the GNC system. Previous experiments demonstrated the robustness against tracking of CCD hot spots by the VBS;
- autonomous closing with high speed in open loop to close in on Tango down to 15km, and subsequent control of perigee distances with final delivery at 50m to an accuracy of 5-6m;
- autonomous station keeping at the delivery point, in the presence of severe measurement errors due to the absence of Close Range camera data, sufficiently accurate to allow AFF to take over control on time tagged command;
- ability of autonomous navigation using only angle measurements, without range, down to a proximity of 50m, with a best performance of about 2m.
The experiment is the culmination of many months of work to get all the bits and pieces together, and a large thanks goes out to all involved, making this happen.
The Mango satellite manoeuvred itself last week out to 30km to start the second Autonomous Rendez Vous campaign. We were pleased to see that also at this distance we still had our intersatellite link holding well. However, we had several problems with among others the Vision Based Sensor, and that led us to move back to 15km distance. From here, an autonomous return was performed by Mango on the optical navigation only to 100m distance, ready for the CNES GNC 2 experiment. This week we started with ARV 2 again, and moved again to 30km distance. Mango arrived there yesterday, and this time the VBS is working fine.
So far the GPS navigation is still working as the intersatellite link is not giving up. Earlier assessments indeed showed that we have enough margin to make it to 30km, but we keep an eye on it. If the link gives up, then Tango will start switching its radio and GPS every 3 minutes until it hears from Mango again. Not much of a problem, but it would be nice to have the capacity to send commands to Tango while we are out there.
All this is a preparation for the ARV2 experiment. From 30km distance, we will attempt to detect Tango with the VBS, and then do an autonomous approach to 30m distance.
Below you find an image that was taken with the VBS from about 10km distance last night.