Prisma Satellites - DLR Experiment: AOK

(Duration of 30 days)

Abstract
The DLR's Autonomous Orbit Keeping (AOK) experiment is a secondary mission objective and is intended to demonstrate autonomous orbit keeping of a single spacecraft (MAIN) on a routine basis. This experiment slot is shared and conducted at the end of the PRISMA mission when only the MAIN spacecraft is operated.

Description
The AOK experiment slot is plit in 2 units:
1) In a first phase AOK is operated in open-loop mode (ca. 5 days). This operational phase is intended to test and verify the basic AOK functionalities, to verify the correct computation of da/dt, to calibrate input parameters and to verify the correct functioning of automatic switch to on-board propagated reference orbit. Due to the fact that no orbit control maneuvers are actually executed, the real Longitude of Ascending Node (LAN) and orbit's semi-major axis will deviate freely from the reference LAN and semi-major axis. An analysis of the free motion will allow to estimate with a good approximation the actual atmospheric drag value and to verify the accuracy of the computed da/dt.
2) After enough confidence has been acquired about the bahaviour of AOK, a second phase is initiated in which AOK operates fully autonomously and in closed loop. This part of the experiment is split in two operational phases. In a first phase (ca. 18 days) AOK uses a reference orbit generated on-ground with the a high accuracy gravitational field model and uploaded to the satellite by telecommand. In a second phase (ca. 8 days), the reference orbit on-board the satellite will be let to expire and AOK will switch to an on-board propagated reference orbit.
The reference orbit to be uploaded is generated on a routine basis at the DLR's PRISMA Experiment Control Center based on the results of the Precise Orbit Determination (POD) verification layer. The reference orbit is generated by an orbit propagator that uses the last POD estimated state vector as initial state for the orbit propagation. The validity period of a reference obit uploaded from ground is three days.

Reports:

Fine control!

As we do not give any news about the AOK experiment since more than one week, one could think that the experiment is ended or something got wrong. Nothing of this! We were silent but not absent and always working hard. After the reference orbit reacquisition phase (Fig. 8) a 5 days control-tuning phase took place. This means that the AOK controller’s degree of freedom has been kept restricted (by limiting the allowed maneuver magnitude and setting long maneuver duty cycles) in order to understand which are the main constraining factor for the control performance. In simpler words, AOK was given the possibility to keep the real orbit close to the reference using only one “shot” in long time periods. In the perfect world this single maneuver opportunity would be exactly the right one to keep the orbit with a minimal fuel expense. In the real world there are elements as the navigation accuracy or the orbital perturbations forces environment which limit the accuracy of this single shot. Thus in the control-tuning phase despite of the degraded control performance I was able to understand a lot in order to give the right control settings to AOK for the best control accuracy performance. We are now since almost two weeks in the fine control phase in which AOK is controlling the LAN with an accuracy of 10 m.

The piece of the action

And now, finally a new issue of our column “The piece of the action”!

Fig. 13 shows better than any explanation what I do intend for fine control. The red lines define the ±10 m control window. As always it is very interesting to observe the difference between what AOK can see on-board and the reality (POD) and how this difference influences the control. It can be noticed how AOK can keep the orbit not with a single shot but with some small corrections after a main orbit maneuver.

Figure 13 Semi-major axis deviation (top), LAN deviation (middle), magnitude of along-track maneuvers (bottom) computed by AOK during the guidance process as between the 5^th and 8^th of August

Monitoring volcanoes Etna and Stromboli…

Now that MANGO is really locked on its reference orbit, DVS data-takes (pictures) can be planned well in advance for long periods of time and with a high accuracy. In fact knowing the reference orbit (stored on my laptop…) now means knowing in advance where the real satellite is at anytime. How does the process of taking a picture with MANGO work? It is really simple.

First I run a little software I wrote to which I give the coordinates of a location on the planet and the period of time in which I would like to take a picture. The program gives me as output a plot showing the different possible tracks I could use in that period of time and when I should switch-on the DVS camera. For example I want to take a picture of volcano Etna that right in the last weeks showed up some activity. I give to the software the Etna’s coordinates and the time period 8^th to 11^th of August. I have to consider that a single DVS data-take means 13 shots taken in 10 minutes over about 4000 km on Earth. Fig. 14 shows one output of my software (red tracks are during the daylight). From the plot it is evident which the ground track to be used is. I set up things in a way that the Etna picture will be about in the middle of the 13 pictures sequence. From this consideration I schedule the start of the data-takes (switch-on of the camera) during orbit 6064 at 16:52:00 UTC on the 10^th of August when MANGO is flying about over the city Riga.

Figure 14 Possible MANGO satellite’s ground tracks to be used for DVS data-takes on volcanoes Etna and Stromboli

And here we are! Thanks to the operations team of GSOC who has made this possible! Figures 15 and 16 are pictures 5 and 6 of the data-take. In Fig. 16 the plumes from volcanoes Etna and Stromboli can be noticed (and admired!).

Figure 15 South Italy as seen by MANGO on the 10^th of August

Figure 16 Volcanoes Etna and Stromboli as seen by MANGO on the 10^th of August

Dreaming of a holiday from 700 km

Looking at Figures 15 and 16 make also dream of an holiday on the beach, as well as Fig. 17 taken on the 4^th of August at around 18:00 UTC over the Côte d’Azur

AOK_figure17_thumb

Figure 17 Côte d’Azur as seen by MANGO on the 10^th of August

Where is TANGO?

And now the most popular columns of this blog Where is TANGO?

TANGO is steadily increasing the along-track separation with MANGO, but is still in view by the inter-satellite link.

Figure 18 Relative position MANGO – TANGO in the RTN orbital frame as on the 9^th of August

Written by

Sergio De Florio

2011-08-16 / 15:59:18

Where is TANGO (August 3rd)

Some of our most enthusiastic readers wrote us declaring themselves literally addicted to our popular column Where is TANGO? Here we are!

TANGO is now at 20 km from MANGO (Fig. 12), but the intersatellite-link keeps working!

Written by

Sergio De Florio

2011-08-04 / 12:47:55

Taking Images of Earth from 700 km altitude

Yesterday a series of new images were taken by Mango, this time while AOK is in closed loop. Here are some nice little details from the very latest pictures, enjoy the landscape!

Lake Powell and the Colorado River as it flows towards Grand Canyon (hidden beneath the clouds in the lower left)

The Great Salt Lake and Desert, Utah

Written by

Niklas Ahlgren

2011-07-28 / 16:28:29

Where is TANGO? (July 26th)

And now our most popular column Where is TANGO?

Figures 10 and 11 show the evolution of the relative position of MANGO with respect to TANGO in the orbital frame RTN from the 18^th to the 26^th of July. The increased relative drift rate in the along-track-direction after the beginning of the 1.5 days reacquisition maneuver can be appreciated in Fig. 10 as well as the return to a smaller drift rate after the execution of the counter-maneuver by AOK. Fig. 11 gives a picture of the situation in the in-orbital-plane and out-of-orbital-plane.

Figure 10 Relative position MANGO – TANGO in the RTN orbital frame as on the 26^th of July

Figure 11 Relative position MANGO – TANGO as on the 26^th of July. In-plane (top) and out-of-plane (bottom)

Written by

Sergio De Florio

2011-07-28 / 16:17:46

The hunt goes on…

25 July

Many of our enthusiastic readers complained that I did not give yet any account of what happened during the week-end. I will try to rebuild their trust with the following detailed account.

When I arrived yesterday (Sunday 24^th) in the control room to check the telemetry, I expected to find that AOK had commanded the counter-maneuver when the LAN deviation was -10 m and thus that the control was in steady-state. But no maneuvers had commanded and the LAN deviation was still diving towards increasing negative values. After a first moment of bafflement, I understood what happened. The software was behaving exactly as it was supposed to do. The problem was a mistake made by me in organizing its activities during the RO reacquisition. I should have studied better the instruction manual…

What happened? Just 30 minutes before the time AOK would have commanded the counter-maneuver, a new block of the reference orbit stored on-board was activated. Now, when this happen AOK deletes its memory (used to compute correctly an orbit maneuver) and it needs then four orbits to reconstruct it. This means that from the moment a new RO block stored on-board is activated, AOK cannot command a maneuver for three orbits. And so it did on Sunday morning. The result was that it was allowed to command the counter-maneuver only when the LAN deviation was already at -60 m. Not too bad, the most important thing is that AOK behaved correctly. I will have to wait more than one day longer for the control to reach the steady state and this has been a good lesson for me to keep in mind that things have to be planned more carefully. However I verified that AOK had brought the LAN deviation to +10 m in exactly 1.5 days as it was ordered to do and with a single shot only!

The piece of the action

Let’s update now our column “The piece of the action” always taking as reference Fig. 1 which is the desired “action”.

Fig. 8 shows the entire RO reacquisition phase, the counter maneuver and another anti-along track reacquisition maneuver commanded by AOK. Fig. 9 shows the on-board navigation accuracy.

Figure 8 Semi-major axis deviation (top), LAN deviation (middle), magnitude of along-track maneuvers (bottom) computed by AOK during the guidance process as on the 25^th of July

Figure 9 On board navigation accuracy – Position in RTN

Written by

Sergio De Florio

2011-07-28 / 16:14:29

Closed-loop! At the hunt for the reference orbit

22 July

This morning I decided to put AOK in closed-loop one day earlier than scheduled as the first four days of commissioning phase have been enough to verify that everything is working properly. As soon as AOK will be put in closed-loop it will start the reference orbit reacquisition phase. This means that it will have to bring the LAN deviation (Figures 1 and 6) back into the control window (±10 m) starting from 300 m. This is one of the most delicate phases of the experiment as a large maneuver is required and the dynamics is much faster than it is in steady-state. The problem was now deciding how long was to be the re-acquisition phase. There is the possibility to tell to AOK by TC a time term by which he has to accomplish the job. Larger will be the given time, slower will be the RO reacquisition and thus smaller the maneuver commanded. Imposing a reacquisition time thus means telling AOK to bring the LAN deviation to +10 m in that time if the LAN deviation is positive. After that, when the LAN deviation diving towards always smaller (negative) values will reach -10 m, AOK will command a counter-maneuver to let the LAN deviation within the control window and thus bringing the control dynamics in steady-state.

The counter maneuver is the important event to be monitored and we will have to be at the control centre for this event. Now, today is Friday…Monday would be too late. We have to decide if we prefer working at the control centre on Saturday or on Sunday…We decide for Sunday. I will impose a reacquisition time such that the counter-maneuver will take place on Sunday morning at around 10 o’clock (local time) that is a reasonable time for us to be there on Sunday morning. It’s like having a date with a maneuver on Sunday morning, but so to say everyone with his own dates…

By the way, it is a strange law of space operations that if something important (expected or unexpected) has to happen, it will be on Sunday and possibly in the early morning!

Then we are in the control room during the 10 minutes passage starting at 11.00 UTC (13.00 local). I’m going to push the red button that will put MANGO’s orbit in AOK’s hands when the red telephone rings: “Stop everything! If AOK will start now commanding maneuvers the collision probability with debris IRIDIUM 33 DEB (ID 35295) will increase!”. The size of this debris is estimated to be around 10 cm and the minimum distance/radial distance foreseen with MANGO is about 518/69 m. (Of course there were neither red buttons nor red telephones, but it was nice to say it…). We finally manage to push the red button in pass 5788 at 12.35 UTC and AOK executed an along-track maneuver of +0.0136 m/s.

Written by

Sergio De Florio

2011-07-28 / 16:07:17

Where is TANGO?

By popular demand from our devoted readers we open today a new column of this blog: Where is TANGO?

Before showing where TANGO is, some technical explanations are due.

From the start of the AOK experiment, TANGO’s on-board GPS receivers are switched-on for only one orbit a day. In fact gaps in TANGO’s GPS data are likely to occur due to the weakening of the inter-satellite link with the always increasing MANGO-TANGO along-track separation and its relatively fast rotation (TANGO is in safe mode). These GPS data gaps can disturb the proper functioning of MANGO’s on-board navigation filter working in the operating mode and with the settings optimal for the AOK experiment. These one-orbit GPS data are anyway sufficient to perform a POD (with a much degraded accuracy of course) and to know with an acceptable accuracy where TANGO is. Fig. 7 shows the evolution of the relative position of MANGO with respect to TANGO in the orbital frame RTN from the 15^th of July when the formation geometry reconfiguration started after I sent an e-mail indicating the desired formation configuration at the start of the AOK experiment. The relative along-track separation on 19^th July was about 3.5 km. The constant drift of TANGO getting away from MANGO can be appreciated in the middle plot.

Figure 7 Relative position MANGO – TANGO in the RTN orbital frame

Written by

Sergio De Florio

2011-07-27 / 12:12:32

What’s going on

At the third day of commissioning phase, we have to point out which is the current situation and to decide if put AOK in closed-loop already tomorrow or on Saturday as scheduled in the present plan.

This is what I could see this morning on the telemetry AOK page (Fig. 3) in the last pass:

Figure 3 AOK telemetry page

From the last TM data, it results that everything is OK! The present LAN deviation is around 238 m and the computed maneuver is around 0.01 m/s.

Now some analyses on the longer period in order to check if there was any anomaly and in case to explain it.

First of all the GPS navigation system has to be checked. I plot the trend of more than 50 TM parameters and everything is OK. There is indeed some small anomaly but it is well clearly explained.

Now let’s take a look to the GPS based on-board navigation accuracy. This check is performed by comparing the position and velocity of the spacecraft estimated on-board with the position and velocity estimated on-ground with the POD process. The orbit determination made on-ground is also based on GPS data but it is far more accurate due also to the availability of better information about the GPS satellites positions and to the far larger computational power available. The navigation accuracy (Figures 4 and 5) is evaluated in local orbital frame (R axis radial, N anti-cross-track and T along-track oriented). In a more “drinkable” (friendly) language, R is along the line joining the centre of mass of the Earth with the centre of mass of the satellite, T is about along the velocity vector of the satellite (this is rigorously true for a perfectly circular orbits) and N is perpendicular to the plane formed by T and R.

The position of MANGO is estimated on-ground with the POD process with an accuracy of the sub-centimetre level!!! We can thus state that the orbit estimated by the POD process represents very well the real orbit.

Figure 4 On board navigation accuracy – Position in RTN

Figure 5 On board navigation accuracy – Velocity in RTN

And now let’s take a look to what AOK sees on-board MANGO. Fig. 6 represents a “piece of the action” of Fig. 1 after almost three day since the beginning of the experiment. The points noted as POD in Fig. 6 represent what is really going on at each ascending node passes (at least with an accuracy of the sub-centimetre level) while the points noted with AOK represent what the AOK controller can actually see on-board the spacecraft (as the position accuracy available on board is around 2 m).

Figure 6 Semi-major axis deviation (top), LAN deviation (middle), magnitude of along-track maneuvers (bottom) computed by AOK during the guidance process

Written by

Sergio De Florio

2011-07-27 / 12:08:02

Taking Images of Earth from 700 km altitude

During the AOK experiment, MANGO will be not any more the deputy satellite of a two spacecraft formation flying mission, but a remote sensing satellite i.e. a satellite dedicated to take images of the Earth.

The on-board Digital Video System (DVS) camera that was used so far many time to picture TANGO, sometimes will be pointed towards the planet below.

Figure 2 New Mexico to Montana, U.S.A. as seen by MANGO on 20^th July 2011

Written by

Sergio De Florio

2011-07-27 / 11:26:28

Let’s start with AOK

The AOK experiment is started. On 18^th July during pass 5729 at about 10:20 UTC AOK was awaked from his long sleep and began to work. In the preceding pass the first block of reference orbit was uploaded to the satellite. As initial state of the RO the position and velocity vector of MANGO at 1.00 on 18^th July as estimated by the GPS based Precise Orbit determination (POD) process has been taken.

These are the relative orbital elements (taking TANGO as the reference satellite) requested by me before the experiment start:

ada = 2 m, adex = 0, adey = -300 m

adix = 0, adiy = -400 m, adu = -3000 m

And just before the start of AOK the real relative orbital elements where:

ada = 2.9 m, adex = -58 m, adey = -272 m

adix = -36 m, adiy = -396 m, adu = -3205 m

Very good job! Many thanks to the personnel of the German Space Operations Center (GSOC) involved.

These initial conditions where meant to have TANGO flying “in front” of MANGO at a safe distance and drifting away thanks to the positive da (TANGO flying “below” and thus faster than MANGO). The relative eccentricity and inclination vector parallel (collision avoidance criteria).

Some tension before receiving the first Telemetry (TM) data from the spacecraft. Sometimes a glance to the Lego model of the PRISMA spacecraft in the control room, thinking that now we are not playing with Lego but with the real stuff…But AOK is alive and working properly! The initial LAN error of the spacecraft with respect to the RO was computed by AOK as about 152 m.

The first five days will be of commissioning phase. This means that AOK will be allowed only to compute orbit correction maneuvers but not to command them to the propulsion system. The commissioning phase is intended to verify that all the software functionalities are working correctly, to evaluate the accuracy of the on-board navigation filter and the magnitude of the main perturbing forces (atmospheric drag and lunar-solar gravity) by the evolution of the deviation of real semi-major axis and LAN from their reference values. In this phase MANGO will be in free (uncontrolled) motion, the LAN deviation will steadily increase and the negative value of the difference between the real and the reference semi-major axis will steadily increase in magnitude due to the decay of the real orbits’ semi-major axis caused by the atmospheric drag.

The reference orbit, propagated once for the entire experiment, will be uploaded to the satellite in blocks every 3 days. Each uploaded block has a validity of about 3.2 days corresponding to the available on-board buffer which contains the GPS time of the first reference LAN, 50 consecutive in time reference LAN values and 50 semi-major axis values at the ascending node. The semi-major axis values are used by the AOK software for atmospheric drag on-board estimation. Each block is generated in a way to assure at least 5 hours time overlap with the contiguous ones. This RO upload strategy allows the exploitation of 3 days of the available 3.2 of each block ensuring at the same time the availability of two consecutive passes for the upload. The possibility of generating blocks with different time overlaps gives a great flexibility in the scheduling and re-scheduling of RO uploads.

Some planning and hopes

The TCs procedures that will be sent to the spacecraft for the AOK experiment are first validated in DLR by means of a tool capable of a realistic simulation scenario involving DLR’s flight software modules. Once it has been verified that the simulation output does not present an unexpected behavior, the TCs procedure is validated in SSC by means of a simulation involving the entire PRISMA flight software. The results are finally double checked at DLR. Fig. 1 shows a representative plot of the validation of the TCs procedures that will be uploaded to the spacecraft during the AOK experiment. The plots represent the first 20 days of the experiment. The deviation of the real semi-major axis and the LAN with respect to the reference and the maneuvers executed by AOK are shown. The red lines represent the LAN control window and the green cross the time at which AOK is put in closed loop. A discontinuity can be noticed in the LAN deviation of Fig. 1 because of the delay in the upload of the new RO during the commissioning phase. The expiration of the RO stored on-board triggers the transition to on-board propagated RO and the LAN deviation starts again from a null value as the on-board propagator uses as initial state the actual satellite state at the moment of the mode change. The same event can be noticed four days before the end of the experiment, but this time in closed-loop. A timed 1.5 days re-acquisition maneuver of 0.0113 m/s is executed at the moment the closed-loop is commanded by TC.

Fig. 1 represents more or less what we hope to see in reality during the AOK experiment!

Figure 1 TCs validation - Semi-major axis deviation (top), LAN deviation (middle), magnitude of along-track maneuvers (bottom) computed by AOK during the guidance process

Written by

Sergio De Florio

2011-07-26 / 15:14:02

AOK - An Introduction

Main goal of the AOK experiment is demonstrating satellite autonomous orbit control using a guidance law for the orbits’ Longitude of Ascending Node (LAN). Using GPS-based absolute navigation data, AOK shall command thrusters activations in the orbital frame to autonomously control the orbit within a predefined window. The main requirement of the experiment is to demonstrate a control accuracy of the osculating ascending node of 10 m (1s). Main differences with respect to similar experiments performed in the past are the extremely strict required control accuracy and the full autonomy enhanced also by the possibility of on-board reference orbit propagation. This technology is foreseen to be widely used in the area of remote sensing missions.

Just a moment, we are already lost in the strange language of scientists and engineers…

What’s LAN control?

OK, let’s make a step back, this is anyway a blog, not an handbook of space flight dynamics. What is LAN control? No wait…what is a LAN?

It is simple. The MANGO satellite is flying at about 700 km and his orbital plane has an inclination of about 97 deg with respect to the Equator. When he makes a complete revolution around the Earth (about 15 revolutions per day), his ground track intersects the Equator two times. The LAN is the longitude on Earth at which the satellite’s ground track intersects the Equator while flying from South to North. Now, why controlling the LAN?

If one wants to observe with a satellite, the same location, for example at the Equator, once per week always in the same day of the week and at the same time, a specialized orbit is required. This orbit, a repeat track orbit, can be designed using the laws of orbital mechanics, but is in fact an ideal trajectory. A real satellite put after launch into this ideal orbit, will soon get away from it because of the perturbation forces (not only gravitational) influencing his real orbit. For example MANGO’s orbit is mainly influenced by the atmospheric drag (Yes man! At 700 km the Earth’s atmosphere has still something to say). Thus for this reason if we want to keep a satellite on an ideal repeat track orbit, we have from time to time to correct his trajectory by means of orbital maneuvers.

OK, now it’s clear. But all this stuff is probably one of the first things that where implemented since the first remote sensing satellites was launched. What’s new with AOK? Why the experiment?

So far, orbit keeping maneuvers have been computed on ground and sent by Telecommands (TCs) to the satellite to be executed by the on-board thrusters.

With AOK is the satellite itself which autonomously estimates when an orbit keeping maneuver is needed and give command to the on-board thrusters to execute the maneuver!!!

How does it work?

The LAN is controlled simply by giving a velocity increment or decrement in the direction of flight.

MANGO always knows its absolute position with the accuracy of 2 m by means of his GPS based on-board navigation system realized by DLR. The ideal reference trajectory is stored on-board the satellite’s mass memory so thus MANGO always knows at which longitude it is supposed to cross the equator. At each revolution he has, so to say, a date at the Equator with an ideal satellite and he knows exactly if he is in time or note. As we do (or better as punctual people do…), when MANGO is in late it hurries up, when he is early he slows down (as anyway he has to get in time there but it will not able to stop there), when he is in time he simply maintain its velocity. To hurry up MANGO has to decrease his semi-major axis with a velocity decrement and vice-versa to slow down. It is not very intuitive, but this is orbital mechanics.

But wasn’t PRISMA a formation flying mission? Where is TANGO?

TANGO is in holiday. The AOK experiment concerns only MANGO as it is an absolute orbit control experiment. Just before the beginning of the experiment TANGO is put in a safe position and will drift away until the end of the experiment. Information from TANGO will be received by MANGO until the along-track separation will be large enough to render impossible the communication between the two satellites.

Written by

Sergio De Florio

2011-07-26 / 15:05:39