GNCDE AS DD&VV ENVIRONMENT FOR ADR MISSIONS GNC

Luigi Strippoli*, Nuno Gomes Paulino*, Julien Peyrard*, Pablo Colmenarejo*, Mariella Graziano*, Jürgen Telaar**

*GMV, Spain, lstrippoli@gmv.es, npaulino@gmv.es , jpeyrard@gmv.es , pcolmenarejo@gmv.es, mgraziano@gmv.es

**Airbus Defense and Space, Germany, juergen.telaar@airbus.com

ABSTRACT

GNCDE ([1]) is an integrated GNC development and verification environment, developed by GMV in the frame of an ESA-GMV co-funded activity. It contains templates of four different adaptable scenarios (Rendezvous and Docking, 3-axis stabilization, Formation Flying, Launchers), a complete set of libraries (based on ESA SPACELAB standard, [2]) for sensors, actuators, DKE and GNC blocks, and a set of tools to ease up the design and analysis of the mission, e.g. guidance trajectories design, control and estimation synthesis, covariance analysis, Monte Carlo campaign, Statistical analysis, Autocoding, 3D visualization through direct connection with tools like Celestia, etc. GNCDE has been already successfully used to design the GNC of different rendezvous missions such as Advanced Re-entry Vehicle ([3]) and Mars Sample Return Orbiter ([4]) and it is the current development environment for the formation flying software of PROBA-3 phase CDE ([5]).

This paper will focus on the utilization of GNCDE for assessing GNC concepts of two different ADR scenarios, both aimed at the post-life disposal of ENVISAT:

1) Design, development, verification and validation of the GNC for RDV and de-orbiting phases of E-Deorbit mission, currently in phase B1. E-Deorbit is so far the most advanced ESA activity with the objective of de-orbiting ENVISAT. It is unique in its operational complexity and requires a high reliable and strongly validated GNC design.

2) Quick preliminary feasibility evaluation from GNC point of view of PRIDE-ISV vehicle used as active debris removal spacecraft. PRIDE-ISV is the ESA program aimed at developing a reusable robotic spacecraft with different in-orbit servicing capabilities, among which the possibility to serve as an ADR vehicle.

The high flexibility of GNCDE has permitted to adapt very quickly the rendezvous and docking template (originally used for an ATV-ISS docking scenario) to the two different ADR scenarios, parametrizing it opportunely

to include configuration, initial orbital and attitude data of both ENVISAT and the chaser spacecraft. Sensors and actuators parameters have been also modified to take into account the typical accuracies and errors in the two cases. The rendezvous trajectories have been tailored to these scenarios and the GNC laws adapted to their specific needs.

In the case of PRIDE-ISV scenario, the study preliminarily indicates that the vehicle could be potentially suitable for an ENVISAT ADR mission. Using the link between GNCDE and Celestia, a video showing the capture phase, including synchronization between PRIDE-ISV and ENVISAT, has been also set up. In the case of E-Deorbit, the work to be done has a longer schedule aiming at a fully validated GNC and the design is still on-going. The paper will present the process which is being followed for GNC DD&VV of this specific scenario, how this process is supported by the GNCDE environment and the available preliminary results.

Index Terms— GNCDE, PRIDE-ISV, E-DEORBIT, ENVISAT, ADR

1. GNCDE OVERVIEW

The GNC design process usually comprises a set of different disciplines and requires involvement of a team composed of people with different background knowledge and formation.

The set of knowledge areas usually involved to a certain extent in the GNC design process includes:

- Mission design and planning- Spacecraft systems knowledge- Trajectory design- Control design- Sensor technology- Navigation strategy and Navigation filters design- Onboard SW coding- SW verification- System verification (including HW in the loop)

The GNC design loop is, in general, an iterative process, where the designer moves forward and backward through the several steps of the sequence, in order to refine the functions being designed. Indeed, initial requirements and assumptions can be usually reconsidered in view of the results obtained from either preliminary analyses or detailed performances evaluation. Due to this iterative nature of the work, the handling of the design data, including inputs and outputs to the process (requirements, synthesis models, parameterization of the models, mathematical representation of the navigation and control functions, etc...) needs to be managed in a coherent way and made available in a suitable shape to every support tool being used in the process.

All these facts point to the need for integrated GNC development environments that not only provide the tools able to support the analysis, synthesis and evaluation activities required but that also manage in an integrated way the data being used within the full process. GNCDE is such a kind of integrated GNC development environment, able to provide these tools and the data coherence mechanisms among them. The next figure summarizes the GNC DD&VV process and the support provided by GNCDE during the whole cycle.

GNC Requirements

Analysis

Analysis & Design Environment preparation

Detailed Performance Assessment Environment

preparation

GNC Detailed Performance Evaluation

GNC Requirements

Satisfied?

A&D Template

(linear models)

Mission Template

(non linear models)

Control Synthesis &

Analysis

Navigation Synthesis &

Analysis

GNC SW coding

Preliminary Guidance Design

Navigation & Control Reqs. Consolidation

Control and Estimator

Design Tool

Guidance Tool

Covariance Analysis Tool

CAD Tool

Test Environment

Model Libraries

MonteCarlo Tool

Statistical Analysis Tool

3D Visualisation

Tool

Model Libraries

AutoCoding Tool

No

Yes

HW in the loop testing

Dynamic Testing

GNCDEResources

GNCDEResources

GNCDesign loop

Legend

Design Process Flow

Tools Contributions

GNCDE Tool or resource

Steps of the GNC design

loop

Equipment Selection & Sizing

Figure 1: GNCDE support along the GNC DD&VV Process

2. ENVISAT OVERVIEW

ENVISAT ([6]) is a European satellite, launched in 2002 for Earth observation purposes. With 10 instruments aboard and at 8 tons, ENVISAT is the largest civilian Earth observation mission ever delivered on orbit. The ENVISAT mission ended on 08 April 2012, following the unexpected loss of contact with the satellite. Since then, ENVISAT has been coasting without any possibility to command it or even communicate with it. It is now considered the top priority target for an ADR mission. ENVISAT is a high priority ADR target because of its mass, its long decay time (at least 150 years) and its location on a Sun-synchronous orbit that is critical due to its proximity to other Earth observation satellites and debris.

ENVISAT’s position is constantly monitored from the ground, along with other debris, to prevent any collision with an active satellite through collision avoidance maneuvers. Currently, ENVISAT orbits at an altitude of 766km, on a nearly circular orbit, with an inclination of 98.346º. While the position and the physical characteristics of ENVISAT, like the mass, center of gravity or inertia tensor, are known with a relatively good precision, it is not the case with its attitude.

When the satellite was lost, ENVISAT started spinning, up to an angular velocity of several degrees per second, converting the satellite into a very fast spinning object (an outlier) with respect to other debris. The spinning period of the satellite has been measured at 135s, with the main rotation around the satellite y-axis. The angular velocity is slowly decreasing with the effect of the perturbations, primarily from the Earth gravity gradient. This paper considers a pessimist conservative value of 5 deg/s for the initial angular velocity, along the y-axis. Figure 2 shows the ENVISAT body reference axes. The high rotation speed and the mismatch between the alignment of the angular velocity and the main moment of inertia causes the target’s attitude to travel to any direction w.r.t. to the angular momentum (Figure 3).

Figure 2: ENVISAT axis

Figure 3: Initial directions of the angular velocity and momentum, where the approaching axis to target

tumbles in the entire possible domain

ENVISAT’s body is made of carbon fibers, thus presenting a high risk of producing debris in case of shock or strong mechanical actions. Therefore, when using a robotic arm, a metallic part of the body must be selected for capture with the robotic arm. One of the most promising locations is the interface ring (origin of the axes in the ENVISAT picture above).

3. E-DEORBIT SCENARIO

The e.Deorbit mission ([7]) is to “Remove a single large ESA-owned Space Debris from the LEO protected zone”. To accomplish this, the mission is designed with a ‘chaser’ satellite launched by a small or medium launcher, which will autonomously perform a rendezvous and docking with the ‘target’ ENVISAT by mean of a robotic arm. The operations include capturing and removing the target from the LEO protected zone accordingly to mitigation rules and mission scenario requirements constrains issued by ESA, which include a robust design of the chaser, able to deal with uncertainties in the target status.

This mission represents a new type of mission with specific needs:

- Meet target satellite and keep a relative position to the target for capture operations

- The target is non-cooperative and non-passivated, tumbling, and the chaser shall stabilize it

- Perform a controlled de-orbiting of the combined target-chaser system

The capture technique employs a robotic arm combined with a clamping or fixation device, selected based on performance, safety, cost, and its verification prior to utilization in the mission.

One of the main critical aspects of the mission is the unknown attitude motion of the target, which puts a heavy demand on the GNC system during rendezvous and capture, where the Chaser spacecraft needs to keep a stationary relative position with respect to the target’s body frame, compensating the centrifugal acceleration while keeping a constant relative attitude. The attitude motion shown in Figure 3 demands a very generic guidance, and has a large impact on the delta-V necessary to achieve the capture point and keep it.

The current chaser wet mass is about 1470 kg, with more than 50% of propellant (~820 kg). 12+12 AOCS thruster of 22 N are foreseen, plus four 220 N and two 425 N thrusters acting as main engines. Camera at far range, LIDAR at medium range and LIDAR 3D at shortest ranges are assumed as relative sensors.

Figure 4: e.Deorbit capture concept explored in Phase B1

3.1. Mission phases and contributions

The rendezvous phase is split into far, medium range and close range (synchronisation) phases. The Stack phase concerns the operations of the composite system, including stabilization, pointing and deorbiting.

GMV provides the guidance and control during the rendezvous sub-phases (excluding the relative navigation):

- Homing (far rendezvous phase), from 8km to 800 m from target

- Safe hold relative orbit at 800 m (an ellipse trajectory never crossing V-bar)

- Closing (medium range rendezvous), from 800 m to 100 m

- Parking hold point, with station keeping over V-bar at 100 m

- Inspection trajectory: from 100 m perform fly-by around the target and back to parking hold point.

- Approach along V-bar to 30m- Fly-by to the direction of target’s angular

momentum vector- Attitude synchronization, and approach to 7 m- Final approach and fly by to capture point in

target’s body frame

Figure 5: Example of trajectory for final approach

3.2. GNCDE utilization

The design and simulations are carried out in the GNCDE, using extensively its pre-validated SpaceLab libraries (both for real world dynamics and OBSW), the ACED Tool for control synthesis/verification and the MonteCarlo engine. Further, through simulations, analyses are carried out, such as consumption evaluation and sensor placement.

Using the GNCDE template framework, two MIL simulators are developed: the RDV template and the STACK template. The former runs the rendezvous phase from far range up to synchronization, while the latter is in charge of simulating the composite body (chaser+target) phases (detumbling and deorbiting).

Figure 6: GNCDE template structure

Each GNCDE template contains an OBSW module which includes the GNC and management modules.

Figure 7: Structure for the On Board SW in the GNCDE template

For an overview, Figure 7 displays the structure adopted in the simulator at this analysis phase:

- Mission and Vehicle Management: it manages the GNC mode transitions and selection/deselection of equipment (sensors and actuators) and GNC functions for each of the GNC modes. A programmable model is already available in GNCDE libraries according to SPACELAB standard, which will be tailored for each simulator.

- Control algorithms: they include the robust control laws for each GNC mode, implemented with ACED Tool according to synthesis/analysis techniques, expressed by means of state-space formulation.

- Navigation algorithms: for STACK it includes the filtering of the absolute sensors while in the RDV template it relies on a performance model of the visual/LIDAR based navigation filter (far range visual camera navigation during the homing, LIDAR performance model at safe hold point).

- Guidance algorithms: they include the generation of the chaser reference relative position and attitude profiles, as well as the computation of the required

feed-forward actions from known spacecraft dynamics. Although some guidance libraries were available in GNCDE, given the novelty of the scenario, new libraries were added and integrated in GNCDE for specific parts of the trajectory such as the spiral approaches, fly-by along the angular momentum, and station keeping in target’s body frame.

- The actuation management computes the allocation of each thruster magnitude with a simplex optimization, based on the commanded torques and forces from control. Also this model was already available in GNCDE/SPACELAB framework.

The control design is carried within the H∞ framework, where the system’s description is combined with the weighting transfer functions to create the augmented plant used in the synthesis. Any set of requirements expressed in frequency through W (s) becomes embedded in the system.

As mentioned, the design and robustness analysis is supported by the Automatic Control and Estimator Design Tool (ACEDTool) which supports the user in the GNC design process by allowing control and estimator design and analysis for linear time invariant (SISO or MIMO and continuous or discrete) systems. The stability and performance objectives are achieved by tuning the weighting functions, and the ACEDTool contains a very generic augmented plant, allowing the expression of requirements on error, output, actuation, and also characterization of generalized inputs, Figure 8. Furthermore, it takes in uncertain systems, and supplies the tools for robustness analysis of the closed loop system, for a first validation of the performance and stability of the controller in the presence of uncertain parameters and unmodelled dynamics.

Figure 8: MIMO robust synthesis and analysis framework, built in the ACEDTool

While for the far and medium rendezvous the control relies on attitude pointing control, for closer operations there is the need for a more agile 6DoF controller to deal with coupled attitude and position relative dynamics, while keeping good tracking in the vicinity of the target. During the STACK phase, a much heavier body implicates a slower dynamics, affected also by the flexible modes of ENVISAT’s solar panel.

Uncertain quantities are considered in the robustness analysis: actuation and sensors (bias, misalignment and delays); mass, inertia and centre of mass (also affected by the robotic arm deployed configuration); and characteristics of the target (masses and also flexible modes).

3.3. Results

This section presents some of the most interesting results obtained so far in the activity. Figure 9 shows the homing approach using a reference trajectory composed of a combination of an imposed drift and ellipse relative orbits. These latter are inherently safe in case of thruster mal function since they never cross V-bar. It also presents the chaser’s tracking trajectory (blue) using the cumulative ΔV profile in Figure 10 to perform the orbital transitions.

Figure 9: Reference (red) and real chaser's trajectory (blue) during homing phase from 8km to 800 m

0 1 2 3 4 5x 10

4

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7V history

m/s

time s

Figure 10: Delta V during the homing phase

The most complex guidance profile is during the final approach and synchronization. Figure 11 details the trajectory results of the chaser w.r.t. to the target’s LVLH expressed in LVLH, where the approach is carried along a slow varying direction of the target’s angular momentum (red). Figure 12 shows the final part of the approach expressed in the target’s body frame, where the chaser tracks the angular momentum direction, performing a wide spiral before flying to a position fixed in target body frame.

Given the type of motion of the target, this is a specific scenario where the thruster is in high demand, with an escalating propellant consumption (Figure 13).

Figure 11: Approach and synchronization trajectory of the chaser in LVLH frame centered at target

Figure 12: Final approach and synchronization trajectory of the chaser in the target’s body frame, along

the angular momentum direction

Figure 13: Consumption during final approach, and synchronization

The flexibility of GNCDE and fast proto-typing using SPACELAB enables other types of analysis, by adding pre-existing libraries to the template. Figure 14 shows for example the comparison of the directions of the star trackers versus the directions of Earth and Sun, during a blinding analysis carried with the GNCDE templates. It resulted that, given the orbit of ENVISAT and the directions of the star trackers, at least a redundant pair can be expected (Figure15).

Figure 14: Star tracker directions during approach before synchronization, versus Earth and Sun

0 200 400 600 800 1000 12000

0.5

1

1.5

2

2.5

3

valid

ity

time [sec]

STR1STR2STR3total

Figure 15: Number of available Star trackers during approach and before synchronization

Using the inbuilt Monte Carlo engine of GNCDE, the MCARLOTool, it has been possible to easily parameterize Montecarlo campaigns. As an example, for design and fitting of the visual based relative navigation filter performance model (to be used in place of the actual visual navigation in fast simulations), relative trajectories were generated with GNCDE for possible ENVISAT attitude motions. The MCARLOTool, with the scattering of the initial angular velocity direction of Figure 16, provided this batch of data which was then used to parameterize the performance model.

-10-5

05

10

-50510

-5

0

5

10

243098

7

46

723337

14

80

84

6356

49

58

12

18

40

89

3

78

6911

66

41

59

47

38

50

77

6116

9

51

62

81

8

65

173985

23

91

31

68

Hbar

7310

Rbar

1132451211109876141318171615192021242322252627283130293332343541403938373642434447464551504948525354565559585763626160666564696867707372717475787776818079828584838687898891909293949596989799100

29

96

25

76

90

70

15

3226

60

2

92

+Xs4

13

97

194543

71

42

Target's angular velocity and momentum in FOT (LVLH)

52

74

95

79

88

34

54

48

5

35

21

86

+Ys

53

28

+Zs

27

3693Vbar

83

6

100

75

55

67

82

57

94

99

64

20224487

Hbar

Vbar

Rbar

Figure 16: Scattered direction of the target's initial angular velocity direction

The same campaign allowed for a first evaluation of the preliminary control performance for final approach and synchronization without navigation in the loop. Some motions of the target are heavier on the control than others, where for a more complex tumbling, the chaser needs to track a more demanding reference. For the batch of 100 different initial directions of the angular velocity of the target, Figure 17 presents the errors distribution at the final capture point (point where the chaser starts capture operations) w.r.t. reference position, velocity and attitude.

0 0.005 0.010

2

4

6

8

10

12

m

Capture point position error

Nr c

ases

0 0.5 1

x 10-3

0

2

4

6

8

10

12

14

m/s

Capture point velocity error

0.04 0.06 0.08 0.10

2

4

6

8

10

12

deg

Capture point attitude error

Figure 17: Preliminary errors with perfect navigation, for the scattered initial angular velocities

These results are just initial design references to have a feeling about the nominal control performances. A

comparison with respect to requirements (on the other hand not yet consolidated at mission level) would not make much sense at this stage, as navigation and uncertainties, to be included next in the loop, will degrade the obtained performances of a factor which is not easily and realistically estimable a priori as it depends on a high number of contributors. With this in mind, it shall be highlighted that these preliminary results are very good and encouraging for the next steps of the activity.

4. PRIDE-ISV SCENARIO

The primary objectives of the PRIDE-ISV project is the definition and development of an affordable reusable European space transportation system to be launched by the ESA VEGA launcher and able to perform experimentation and demonstration of multiple future application missions in low Earth orbit, benefiting to the maximum extent possible from existing technologies, and addressing where relevant progressive technological challenges with limited risks and minimal financial efforts for Europe. Furthermore, the PRIDE-ISV project has to support the innovation of re-entry technologies for reusable systems.

Therefore, in line with the above, the PRIDE-ISV system shall provide routine access and return from orbit, with the purpose to perform in-orbit operation, experimentation and demonstration for, without being limited to:

- Exploration technologies;- Orbital infrastructures servicing technologies;- Earth observation technologies;- Earth science technologies;- Telecommunication technologies;- Micro-gravity experimentation;- ISS cargo return;- Rendezvous and docking technologies

Figure 18: PRIDE-ISV preliminary conceptual design

PRIDE conceptual design is just started and no consolidated data is still available. In the frame of the current analysis the following plausible data have been used, according to the information available from the CDF sessions held at ESTEC ([8]):

- Wet Mass: ~2500 kg (~200 kg of propellant)- Inertia: ~[5000 0 -250; 0 15000 0;-250 0 18000]

kg.m2

- 12+12 90N thrusters with ~240s of specific impulse

- Assumed available sensors (with typical accuracies from current state of the art):

o Ground trackingo GPSo Accelerometerso Star Trackerso Gyroso Lidar/Camera

In this activity it has been performed a preliminary feasibility evaluation from GNC point of view of PRIDE vehicle used as active debris removal spacecraft for ENVISAT post-life disposal.

The work, still on-going, has been focused on the final approach and synchronization phase, the most demanding one (before contact) from GNC point of view due to ENVISAT angular motion.

An initial approach is commanded from 50 m to 25m, in the LVLH frame (no need to follow the ENVISAT angular motion during this first phase of the approach). To permit a successful capture the subsequent approach must be made in a reference frame linked to ENVISAT body frame, so that the grapping point is correctly followed. A two-point transfer is needed to reach the desired position in the ENVISAT body frame (25 m away, in the y body frame direction). The PRIDE-ISV synchronizes its attitude with ENVISAT, so that the roof hatch keeps facing grapping point, while keeping a fixed position in the ENVISAT body frame. Next, the PRIDE-ISV approaches ENVISAT along the y-axis till the capture point at 5 m, while keeping the attitude synchronized with the target. Once at the capture point, the PRIDE-ISV performs a station keeping (assumed lasting about 15 minutes), maintaining both its relative position and relative attitude with respect to ENVISAT, in order to deploy the robotic arm in a “relative motionless” environment and perform the capture. The whole trajectory of this phase is shown in the following figure, with a zoom to the station keeping part in Figure 20.

-20

-10

0

10-20-15-10-505101520

-5

0

5

10

15

20

relpos LVLH

y (m)

x (m)

z (m

)

transfer to enBFattitude syncapproach along ySK along ytransfer to capture pointSK in capture point

Figure 19: Approach to capture point trajectory

-10-50510

-10

0

10

-10

-5

0

5

10

x

Pride Final position

y

z

Figure 20: Station keeping in capture point

The previous trajectories have been designed with the support of GATO tool, and a preliminary simple controller and estimator has been synthetized with the aid of ACED Tool. A V of about 130 m/s has been estimated for this approach phase, already taking into account a typical margin of 100% due to GNC, thruster configuration and complete real world in the loop. A representative complete simulation in the rendezvous mission template though has been left among the future work tasks.

Thanks to the link between GNCDE and a visualization tool like Celestia it has been possible exporting the trajectory and attitude data, generating interesting videos of

the approach and synchronization phase (Figure 21) helping to understand if the approach was going as expected.

Figure 21: Approach of PRIDE-ISV to ENVISAT

From GNC point of view, with the assumed sensors and actuators, the PRIDE-ISV seems suitable for carrying out the rendezvous mission to ENVISAT, at least up to the capture point station keeping. The stack phase has been only qualitatively analyzed in this activity, as a detailed GNC design for stack operations (capture, detumbling and de-orbiting) was outside of the scope. The mass of the stack is around 10 tons. The PRIDE-ISV thrusters’ capability, of about 500 to 900 Nm depending on the direction, would potentially allow a detumbling in few minutes. As future work, and when the PRIDE-ISV configuration will be more consolidated, a more detailed study could be done also for the stack phase.

In any case it can be already concluded that, if PRIDE-ISV wants to be used to de-orbit ENVISAT, the vehicle shall be specifically designed around this worst case scenario, both from the point of view of needed propellant and from the point of view of needed AOCS capabilities and equipment.

5. CONCLUSIONS AND FUTURE WORK

GNCDE is a powerful and flexible GNC DD&VV environment already used for developing the GNC of various RVD and formation flying missions at different levels (from preliminary studies up to CDE phase). This paper has presented the utilization of GNCDE in case of two different ADR scenarios for ENVISAT post-life disposal. Thanks to the high flexibility of this environment it has been possible to adapt very quickly the simulation template, parametrizing it opportunely to include configuration,

equipment data, initial orbital and attitude data of both ENVISAT and the chaser spacecraft. The tools provided with GNCDE have been very useful to support the analysis and the designs, including trajectory generation, control synthesis/verification, Montecarlo simulations and smart visualizations.

In the first scenario, the on-going work in e.Deorbit phase B1 activities has been presented. Here, the already validated SPACELAB libraries have been particularly useful to save development time and to avoid wheel reinvention. The ACED Tool is permitting to synthetize the different controllers needed along the mission and validating them with respect to robustness and stability requirements. The Montecarlo Tool allows an easy setup of Montecarlo campaigns for performance evaluation of the proposed GNC. At the time to write this paper, a very good behavior of the nominal control has been observed during the complex synchronization phase, although this does not permit yet to draw conclusions about the final GNC performances wrt. requirements. The activity is still on-going and more consolidated results are expected soon, in line with the project schedule.

In the second scenario, PRIDE-ISV vehicle operational frame has been imagined encompassing ADR scenarios and a quick preliminary assessment has been carried out to see whether the vehicle would be potentially suitable for deorbiting ENVISAT. Though only partial and preliminary analysis have been carried out so far, the first outcomes indicate that PRIDE-ISV could be able, from a technical point of view, to cover that task. In that case though, it is highlighted how the vehicle should be designed around this worst case scenario, due to the AOCS capabilities needed and to the estimated high propellant consumption. Once the definition of PRIDE-ISV will be more consolidated it could be possible to refine the analysis and to cover with more detail all the phases of the de-orbiting mission.

5. REFERENCES

[1] F. Gandía et. al., “GNCDE: an integrated GNC development environment for attitude and orbit control systems”, ICATT 2012

[2] SpaceLab: A Spacecraft Simulation Laboratory Under Matlab/Simulink. A Guide for Users and Suppliers

[3] L. Strippoli et. al., “Mission Analysis and GNC design for RVD phase of ARV mission”, EUCASS, July 2011, St. Petersburg

[4] L. Strippoli et. al., “Integrated vision-based GNC for Autonomous Rendezvous and Capture around Mars”, ESA-GNC, June 2014, Oporto

[5] PROBA3 Mission facts, https://directory.eoportal.org/web/eoportal/satellite-missions/p/proba-3

[6] ENVISAT Mission Facts, https://earth.esa.int/web/guest/missions/esa-operational-eo-missions/envisat

[7] e.Deorbit Mission Requirements Document, GSP-MRD-e.Deorbit, ESTEC, Issue 1, rev.0, 15-07-2013

[8] PRIDE-ISV CDF session 7, ESTEC, May 2015