mission concept and autonomy considerations for ...

30 downloads 227 Views 1MB Size Report
Space Wire Router. Main On-Board. Computer. Payload OBC. Heater. Radiator. LRMLI. MLI. Solar Generator. Solar Generator. Solar Panels. Li-Ionen Batterien.
66th International Astronautical Congress, Jerusalem, Israel. Copyright ©2015 by the International Astronautical Federation. All rights reserved.

IAC-15-D1.4.9

MISSION CONCEPT AND AUTONOMY CONSIDERATIONS FOR ACTIVE DEBRIS REMOVAL Susanne Peters Universität der Bundeswehr München, Institute for Space Technology and Space Applications (ISTA), Germany, [email protected] Christoph Pirzkallt, Hauke Fiedler*, Roger Förstnert t *

Universität der Bundeswehr München, ISTA, Germany, [email protected], [email protected]

Deutsches Zentrum für Luft- und Raumfahrt (DLR), German Space Operations Center (GSOC), Germany, [email protected]

Abstract Over the last 60 years, Space Debris has accumulated to one of the main challenges for the safe operation of satellites in low Earth orbit. To address this threat, guidelines that include a limited debris release during normal operations, minimization of the potential for on-orbit break-ups and post mission disposal have started to be implemented. However, for the long-term, the amount of debris will still increase due to fragments created by collisions of objects in space. The active removal of space debris of at least five large objects per years is therefore recommended, but not yet implemented in those guidelines. Even though various technical concepts have been developed over the last years, the question on how to make them reliable and safe or how to finance such mission has not been answered. This paper addresses the first two topics. With Space Debris representing an uncooperative and possibly tumbling target, close proximity becomes absolutely critical, especially with an uninterrupted connection to the ground station not ensured. This paper therefore defines firstly a mission to remove at least five large objects and secondly introduces a preliminary autonomy concept fitted for this mission. I. INTRODUCTION The idea of implementing autonomy in spacecraft has been followed by some time and is on some level successfully tested for deep space missions and planetary rover. Different kinds of applications combine a limited timeframe for connecting with ground control and unknown parameters about the environment. These features make it difficult to operate a rover and / or spacecraft for the mission time available. Autonomy and on-board processing within a spacecraft make it possible to improve the mission’s data-collection by extending the execution of pre-planned, ground-defined mission operations, expand the available range in objectives and time and limit workload on ground. When performing active debris removal, advantages arising from the implementation of autonomy alter slightly, as for instance the close proximity to an uncooperative target needs special attention regarding fast reaction time to a changed working environment. Not the extension of the mission has the priority but the safety of the operating spacecraft and the target. Addressing the topic of autonomous active space debris removal, autonomy requirements for such mission have to be defined. They again will build the basis for specifications of high-level on-board procedures.

IAC-15-D1.4.9

The requirements stated in this paper are based on the concept of a flexible arm to grab and stabilize a tumbling target. To face unforeseen events or failures, the necessary berthing maneuver needs the capability of goal-oriented mission re-planning. As the close proximity combined with drifting of the objects might end in a collision, the process of switching into safe mode is not an option. Safe mode in this context refers to the procedure coming into operation in case of an unknown failure and results into ceasing all activities until the failure has been resolved by the ground station. To work around the safe mode, advanced failure detection, isolation and recovery concepts need to be involved with the spacecraft able to operate and re-plan by itself. To provide a starting point for such high-level autonomy, the mission had to be defined first - this paper therefore starts with a concept for active space debris removal, its mission architecture and preliminary spacecraft design. Further on, requirements for the autonomy aimed to be implemented result from this setup and are stated in the following. The last part presents a possible approach on how to realize such high level autonomy for the designed mission, introducing a concept used successfully for unmanned aerial vehicles in former tests.

Page 1 of 12

66th International Astronautical Congress, Jerusalem, Israel. Copyright ©2015 by the International Astronautical Federation. All rights reserved.

II. ACTIVE DEBRIS REMOVAL II.I.

Space Debris

The awareness for the threat of space debris to operating satellites and a sustained space environment has increased with the threat itself. More launches, collisions, in-orbit break-ups, or natural decay lead to a growing space occupied by debris, which again increases the collision probability between objects. Depending on size, angle, relative velocity etc. of the impacting debris, satellites being hit can lose their functionality or may be fragmented, adding even more objects to the debris account. Mitigation Guidelines developed by e.g. the Inter-Agency Space Debris Coordination Committee (IADC) for a safer operation of spacecraft and launches would result, if applied, in the reduction of growth of debris. These guidelines are supported by most of the space fairing nations, but not legally binding. Additionally, the guidelines include a limited debris release during normal operations, minimize the potential for on-orbit break-ups and address post mission disposal1. Furthermore, better prediction models have been developed to track the objects and predict collisions with higher accuracy. A warning system gives satellite operators the possibility to move their objects, as far as the spacecraft has this capability. However, these are short-term solutions; ultimately, active space debris removal (ADR) will be necessary to sustain the space environment in the long run as the number of collision generated debris is about to overtake the debris generating due to fragmentations. This again will result in a rising amount of small objects - if the sources are not removed. Analyses of the publically available data of about 16,800 objects – there are about 7,000 more objects that are either not reliably tracked or military satellites and thus do not appear in the report – reveal the low Earth orbit (LEO) as most occupied region, cf. Figure 1. The geostationary orbit (GEO), the second most occupied region, can be described by a more tube-like shape. Due to the objects being concentrated on a smaller area than in the LEO region, it shows a relatively high object density when only considering the enclosed area. With GEO being very important for local observation, customers (and thus funding opportunities) might be more interested to invest in active debris removal within this area. However, directly compared with same altitude and inclination bins, as performed in Figure 1, a smaller over-all distribution is revealed. By first applying ADR in LEO, a reliable technique to safely remove objects can be found in a more cost-effective way. A transfer to higher orbits and by such serving the customers in GEO, can be performed after the successful implementation in LEO and thus in a later stage.

IAC-15-D1.4.9

Figure 1: Distribution of the different object types up to the geostationary orbit. Inclination bins are set to 5°, mean altitude bins to 1,000 km. II.II.

Mission Motivation

Space Debris can be generated in different ways, either by fragmentation, explosion, degradation due to the harsh environment, mission related reasons, collision or simply by reaching a satellites end-of-life without disposal measurements. Until today, debris larger than 10 cm due to fragmentations are the biggest contributor, before collision related debris. The result are collisions at orbital velocities (relative velocities may reach up to approximately 14 km/s), creating even more and smaller debris. The amount of space debris generated by collisions is already that far developed, that it supersedes the amount of space debris created by explosions or environment related reasons. The effect is a cascade effect that will be slowed down or stopped by actual intervention and removal of the source. Even though the main threat2 to operational spacecraft nowadays are fragments from the size of 5 mm to 1 cm, long-term objectives need to concentrate on an overall stabilization of the space environment – by removing objects that are capable to create large amount of debris and are thus called the main driver for the population growth. These objects are satellites or rocket bodies with high masses of 1 t and more. To choose among the high number of object fulfilling this requirement, their collision probability is as well part of the target identification process. Taking into account the simulation and recommendations given by the IADC, the active removal of at least five large objects per year is desired to sustain the known Earth space environment. Even though this number is somewhat notional with assumptions like an immediate removal of the objects from the environment or a repeated launch cycle from the past eight years, it is at least a ballpark figure. When considering former re-entries3, a complete burn out during de-orbit cannot be guaranteed. As a result a controlled reentry should be preferred, the landing area preferable on inhabitant land or in the oceans. The combination of the desire to remove at least

Page 2 of 12

66th International Astronautical Congress, Jerusalem, Israel. Copyright ©2015 by the International Astronautical Federation. All rights reserved.

five objects, and a controlled reentry leads to the idea of having five de-orbiting devices, as well called de-orbit kits, and a main satellite in one launcher. The devices will be attached to the target by a flexible arm, connected to and controlled by the main satellite. The main satellite will not stay with the set-up but guide the other kits to their designated target, the target and the attached kit will be de-orbited together. With the deorbiting device being lost during reentry, the technical complexity is concentrated in the main satellite as it will have to coordinate the berthing and stabilizing phases. By using one launch per year, time and cost of the whole clean-up process can be minimized and an effective measure for the coming years can be found. II.III.

Concept

With the main idea of having a one-launcher set-up with one prime satellite that incorporates most of the complexity when it comes to rendezvous, berthing and stabilization, and multiple devices to de-orbit in a controlled way together with a target heavier than 1 t, a more detailed concept can be developed. Different concepts for capturing an uncontrolled, large object exist. Due to heritage reasons, a high technology readiness level and its feasibility for the mission, a robotic grabbing arm shall be used for further considerations. Examples of such arms can be derived from DEOS4, SDMR5, FREND 36, RANGER 8 DOF7 or OTV8. No specific arm will be set for this mission, however, mass and power requirements follow the DEOS design. Due to the close approach to an uncooperative target, the operation of close proximity becomes absolutely critical, especially with an uninterrupted connection to the ground station and thus constant data exchange not ensured. To solve this problem, high-level autonomy with goal-oriented mission re-planning capabilities shall be implemented. A possible autonomy concept probably adaptable for this mission is presented later in this paper. Rendezvous and docking requires specific flexibility and agility of the berthing spacecraft. Accordingly, the designed spacecraft, as well called chaser, namely ADReS-A for Active Debris Removal Satellite #A, will not carry all de-obit devices with it for the whole mission time, but pick up one kit after another from a parking orbit to shuttle them separately to their designated target. Figure 2 displays the set-up of a) launching into the parking orbit, positioned approximately 30 km lower than the targets orbit, b) ADReS-A shuttling between the orbits carrying one kit, c) ADReS-A attaching the kit to the target, and d) the kit re-entering with the target while e) the chaser taking the next kit to its destination. Not including the main satellite in the cleaning process by e.g. de-orbiting with the last of the five debris objects, leaves the opportunity to refill ADReS-A with additional propulsion and send

IAC-15-D1.4.9

more de-orbit kits into space for further removal. If the decision for a follow-on mission is not made, ADReS-A will however attach itself to another space debris object, then to be assigned, and re-enter with it. c) b)

e) d)

a)

Figure 2: Principle mission architecture, following9. Launching requirements combine a multiple burn capability to reach a circular orbit and the lifting of about 3.1 t mass into an orbit of 940 km altitude. Preliminary analyses lead to Falcon 9 for a suitable launcher. II.IV.

Target

To save mass and consequently limit the costs of the mission, the removed objects should be as close as possible to each other resulting with the shuttling between them requiring a minimum on ∆v and thus propulsion mass. Analysis performed in an earlier phase of this study concentrated on the mass and collision probability of objects listed in the satellite catalogue, adding the requirement of a 2° range for inclination and right ascension of ascending node (RAAN) distribution to search for ‘cluster’10. The investigation resulted in SL-8 rocket bodies at a circular orbit of about 970 km altitude and an inclination of about 83°. Their orbital lifetime without removal is calculated to more than 200 years2. In total, 143 SL-8 R/Bs can be found in this area, with clusters of up to seven objects. With their geometry being similar, the design of the kits does not need to be adapted to different objects so the catching mechanism and strategy can be adapted to one body and transferred to the others. SL-8 R/Bs shall therefore provide the kind of target the removal satellite will be designed for. II.V.

Architecture

The one-year timeframe for the deorbit of five objects does hardly influence the over-all performance. Driving margins for the design of ADReS-A and the kits are feasibility, safety, costs and mass requirements. After introducing the preliminary design of ADReS-A the different application of the sensors during the various phases will be presented, followed by the preliminary design of one de-orbit device. The preliminary estimation regarding the mass-budget and the ∆v-budget, performed in an earlier phase of this study9 is further on updated due to upgraded simulation results.

Page 3 of 12

66th International Astronautical Congress, Jerusalem, Israel. Copyright ©2015 by the International Astronautical Federation. All rights reserved.

Star Tracker Star StarTracker Tracker B

IMU IMU IMU IMU

Main On-Board Computer

Coarse CoarseSun Sun Sensor Sensor

GPS Receiver A

Robotic Arm

Memory Unit PPU

PPU

Li-Ionen Batterien Li-Ionen Batterien Li-Ionen Batterien Li-Ionen Li-IonenBatterien (VL 48E)

Linear Roboter Solar Generator Solar Generator Solar Panels

B

Space Wire Router

Visual Light Camera A

Reaction Wheels PPU

PPU

B

Time of Flight Camera A

Payload OBC Memory Unit

Li-Ionen Batterien Li-Ionen Batterien Li-Ionen Batterien Li-Ionen Li-Ionen Batterien (VES 140)

He Primary Structure N2H4 He

21

21

21

21

21

21

21

21

Radiator

Heater MLI

LRMLI 21

21

21

21

21

21

21

21

Parabol

Parabol

Omnidirectional Omnidirectional

Omnidirectional Omnidirectional

Figure 3: ADReS-A functional architecture ADReS-A Preliminary Design The design of ADReS-A is created around the intention to use a robotic arm for retrieval and the objective to remove the target controlled and immediate. The latter objective is required as rocket bodies due to their architecture most probably will not fully burn up during reentry and thus be a potential threat for inhabited areas. By comparing different space robotic arms, the DEOS4 robotic arm was chosen as role model, however, it will need some adjustments to be suitable for the ADReS-A missions as more flexibility might be required. To attach the kit while handling the SL-8 R/B, a second arm, which is more a linear arm (LRA), capable of moving either forward or backward, shall be used. In such way, the flexible robotic arm (FRA) keeps the rocket body in some distance to the ADReS-A body and the LRA can push the kit in its position, where it attaches itself with a connecting mechanism. Communication needed during the mission, as solely the berthing part shall be fully autonomous, will be performed by one parabolic and two omnidirectional antennas. the propulsion system relies on hydrazine and multiple thrusters as displayed in the functional architecture (cf. Figure 3). The power system relies on body mounted solar panels and two different Li-ion batteries. One version supports the nominal phase, when some sensors and the arms do not need to be active, the additional one the operational phase of the rendezvous and docking maneuver as this requires all systems to work on full power. The on-board computer will be separated into one responsible for the subsystems of the spacecraft and the second handling the robotic arm and

IAC-15-D1.4.9

the camera signals. The distribution allows for separating the tasks locally, however, data transfer is required as the camera data gives feedback to relative distances and probable violations to safety ranges. While the functional architecture can be found in Figure 3, the actual design will be revealed in a later stage of the project. Mission Phases From the mission concept, different phases can be identified that have variant needs. Being in an altitude of about 940 km, the first rough distinction can be made into an eclipse phase and an illumination phase. During the latter, batteries will be charged through the solar panels. The Global Positioning System (GPS) determines the absolute position, while Coarse Sun Sensors (CSS) data and the voltage variation of the solar panels for validation calculate the attitude. Inertial Measurement Units (IMU) provide the translation and rotation determination, reaction wheels are used for adjustments. During the eclipse, Star Tracker (ST) will support the system. In case the reaction wheels need to be desaturated, this will happen during the eclipse phase, the batteries of the nominal phase provide power for the spacecraft to stay operational. The change from parking orbit to target orbit will be performed based on groundstation data as radar can provide an accuracy of the target position of a few hundred meter11. The thrusters are used for this maneuver, which will implement at least two separate burns to reach the new orbit.

Page 4 of 12

66th International Astronautical Congress, Jerusalem, Israel. Copyright ©2015 by the International Astronautical Federation. All rights reserved.

Radiator

MLI-1

Heater

MLI-2

N2

Attach Mechanism

21 x Li-Ion (VES 16) Solar Generator Solar Generator Solar Panels

Louvre LOX

LOX

LCH4

LCH4 Primary Structure

7 x Li-Ion (VL 48E)

Star Tracker Earth Sensor Coarse CoarseSun SunSensor Sensor Antenna Antenna

Transponder

IMU IMU

GPS

On-Board On-Board Computer Computer

Figure 4: De-Orbiting Device functional architecture For the tracking of the target and the approach to about 25 m distance behind, data from the Visual Light Camera (VLC) will be used. The VLC works similar to a star tracker, on-board calculations can determine the object in distances up to a few kilometers. By getting closer in small steps, the increased number of pixel can be used to determine the actual distance, since it has knowledge about the actual size of the rocket body. As decent illumination of the target is required, its operation during sun-phase is recommended. Additionally, the camera should avoid to have the Sun in the Field of View of 20-30°, when turned away from the sun, the reflection of the target must not be too intensive12. This limits the observation and operation time to about 50 min per orbit, resulting in at least two orbits for one step of the close approach. Once ADReS-A has reached a distance of about 25 m to the target, the Time of Flight Camera (ToFC) will supersede the VLC. The ToFC works with infrared light, measuring the time the light needs to reach the target and be reflected. With this procedure, the camera is able to determine the tumbling and axis of rotation, as the algorithm behind focuses on the geometry of a cylinder to fit the target’s movement. The identification will take about 60 min of data acquisition and data evaluation12. The distance to the target has to be limited to 11 m to detect the position of the nozzle and thus determine the actual attitude of the SL-8 R/B, as the cylindrical geometry used so far for approximation reasons has two basal areas of the same kind, leading to an ambiguous conclusion. With the knowledge of actual distance, tumbling mode and attitude of target and spacecraft, the target approach can be initiated. Section III.IV will describe the approach and a possible abort in more detail.

IAC-15-D1.4.9

De-Orbit Kit Preliminary Design The design of the de-orbit device is based on the SL8 R/B nozzle geometry and desire to implement ‘green propellant’. By utilizing namely liquid oxygen (LOX) and liquid methane (LCH4) for the de-orbiting maneuver, possible remains from the re-entry will not contaminate the nature and thus be more environmental friendly. The de-orbiting device is a promising object to broaden the knowledge about the propellants performance in space. ‘Green propellant’ however is not used for ADReS-A as the cryogenic fuels require exceedingly low temperatures to be stored in a liquid state. The additional thermal insulation and storage requirements would add even more complexity to the satellite. As the concentration for ADReS-A is focused on its autonomous performance, well-proven hydrazine will be implemented for the chaser. The de-orbit kit’s preliminary functional architecture is given in Figure 4, revealing the different subsystems and devices planned to be implemented in the kit. The nominal phase for this architecture represents the de-orbit kit waiting in the parking orbit for its deployment. The operational phase in turn is active either with the device rotating itself in the right attitude in the parking orbit to have the right attitude and sun illumination for the solar panels able to charge the batteries to be prepared for the eclipse phase, or with the kit attached to the target nozzle. Additional batteries will be cut in during the operational phase as it requires more operating sensors or the attach mechanism in use. When handling cryogenic propellant, specific thermal insulation is required as the temperature range in which the electronics operate is much higher than the one the propellants need to be stored at to stay liquid and thus usable for the thruster. The device is therefore separated

Page 5 of 12

66th International Astronautical Congress, Jerusalem, Israel. Copyright ©2015 by the International Astronautical Federation. All rights reserved.

into a lower part, which contains the propellant tanks, and an upper part, incorporating the electronic devices. The separation is realized by thick thermal insulation, with louvers for thermal control on the upper part and radiators on the lower part. With the stabilization of the target in combination with the kit performed by spinstabilization, cold-gas thrusters, using the same gas needed to keep the LOX/LCH4 tanks pressured, are implemented. The sensors displayed in the AOCS section are mainly required to keep the de-orbit device in the right attitude in the parking orbit, waiting for its deployment. They will moreover be used to provide a controlled object to attach to, giving feedback on attitude and movement for ADReS-A to smooth the rendezvous and docking process. Communication with ground station to observe housekeeping data will be enabled through patch antennas. As for the main satellite, the actual design will be revealed in a later stage of the project. Mass-Budget The following table gives an overview of the mass distribution of one kit and ADReS-A. Considering the launch of at least five de-orbit devices, propellant and the main satellite, about 3.1 t have to be lifted into a circular orbit of 940 km altitude. Subsystem Structure Propulsion AOCS Thermal Data Handling Communications Energy Dry mass Contingency 15% Dry mass w/ cont Propellant Total mass

Kit Mass [kg] 70.96 80.08 9.53 9.28 3.00 2.26 12.60 187.71 28.16 215.87 198.68 414.55

ADReS-A Mass [kg] 315.30 72.05 53.22 39.66 46.00 6.00 46.21 578.44 86.77 665.21 360.10 1025.31

Table 1: Preliminary Mass-Budget for De-Orbit Kit and ADReS-A

∆v-Budget The ∆v-budget for ADReS-A shuttling between the parking orbit and the designated target was simulated using the General Mission Analysis Tool (GMAT)13, an open-source space mission design tool, supported by NASA. The optimization strategy14 took into account the changing RAAN of the rocket bodies over time, which made a predesigned order of approach necessary. Secular disturbances and the whole mission duration of one year are taken into account as this performance

IAC-15-D1.4.9

allows to spend time on lower or higher orbits to overhaul the targeted object. Without using these orbits, the full approach would have to be perfromed by using propellant to decelerate and accelerate, one of the resources that needs to be saved in space missions. The designed simulation performs phasing maneuvers between parking orbit and target orbit and approximates rendezvous-maneuver of the chaser with kit and target. The time needed to track and dock with an object is fixed by the designer. The simulation shows, that a ∆v of about 400 m/s for the whole mission is required. It incorporates all orbit, inclination and RAAN changes as well as docking and rendezvous maneuvers, resulting in about 190 kg hydrazine for ADReS-A. As rendezvous and berthing maneuvers with an uncooperative target have not been performed so far, it is not fully clarified, how much propellant the maneuver will actually require. The performed simulation therefore follows calculations performed by Udrea and Nayak15. Moreover, a post-mission de-orbit of the satellite itself needs to be implemented. The conclusion drawn from the simulation assume a total consumption of about 360 kg hydrazine, which includes a margin of about 25%. The ∆v-budget for the De-Orbit Kit results from an additional simulation22, based on a finite burn of about 20 min. With the targets mass of about 1.4 t, in addition with the kit about 2 t have to be de-orbited. The maneuver will transfer the set-up from 970 km to approximately 80 km, where a breakup can be assumed due to former observations3. Adding a margin of 20% and known errors as e.g. the tanking error, the required propellant (LOX + LCH4) adds up to about 200 kg. The spin stabilization will be performed using the nitrogen available (cf. Figure 4), requiring about 18 kg of the gas. II.VI.

Space Politics & Space Law

The intention to perform active debris removal brings up major legal and policy issues. The outer space treaty of 1967 states, that the launching State is liable for its space objects and retains jurisdiction while in outer space16,17. This leads to the conclusion that no third party is allowed to remove an objects without former consultation with the launching state. Having objects targeted within the presented mission architecture, that were former owned by the Soviet Union, now by the Commonwealth of Independent States (CIS), requires proper legal and policy framework which is still under development. Questions on e.g. liability in case of damage in space and the possibly resulting new objects or accidents during or after re-entry which may include people and / or property have to be discussed by jurists. As long as no answers have been found to those and similar issues, the

Page 6 of 12

66th International Astronautical Congress, Jerusalem, Israel. Copyright ©2015 by the International Astronautical Federation. All rights reserved.

technology of ADR will need to be tested on objects owned by the removing state. This discussion, however, is out of the scope of this paper. Another challenge is the dual-use nature of ADR. With spacecraft being used by a number of states for nation security purposes, ADR operations might cause mistrust since not all of those states have the capabilities to follow space operations by themselves. Transparency of such activities could be one solution, which again raises the question of intellectual property rights18. The advantage of targeting rocket bodies for an initial removal is their relatively simple shape compared to the various shapes of satellites. Moreover, rocket bodies should not carry any sensitive instruments, which might make it easier to achieve an international agreement for such a removal mission. III. AUTONOMY-REQUIREMENTS III.I.

Motivation

Within the last years, autonomous processes slowly find their way into actual space missions. Autonomy and automation are often used synonymously as they both refer to processes that may be executed independently from start to finish without human intervention. Automated processes, however, “replace manual processes with software/hardware ones, following a step-by-step that may still include human participation. Autonomous processes, on the other hand, have the more ambitious goal of emulating human processes rather than simply replacing them.” This distinction, phrased by Truskowski et. al.19, will be followed within this context. Implemented first on rovers to increase the scientific data returned from missions to other planets with time delays involved, and later expanded to deep-space missions for similar reasons, a high-level autonomy still lacks for low-Earth orbit regions20. Unmanned spacecraft tend to improve risk and feasibility factors for missions. The space proven implementation of fault detection and isolation (FDI) concepts transitions into safe mode as an ultimate reactions to failures which made spacecraft more reliable than ever. The intention to perform active debris removal missions, however, requires more self-awareness and self-operational capabilities of the spacecraft than implemented nowadays. With the ADReS-A mission concept and architecture developed, the requirements for the removal of large debris – in this case SL-8 rocket bodies - can be established. The autonomy for ADReS-A, as mentioned before, needs to provide the capability of the spacecraft, to react in the critical situation of a berthing attempt. Traditional failure detection, isolation and recovery (FDIR) concepts either react to predefined events with a recovery routine, implemented during design time or,

IAC-15-D1.4.9

when confronted with unresolvable failures, put the spacecraft into safe mode. By switching into safe mode, all activities are set to hold and the failure has to be recovered from by ground control. For ADR the objects are too close to use this procedure as, in the case of a failure, their drift might have them collide and thus damage the system, creating what was supposed to be removed – space debris. III.II.

Requirements for ADReS-A

When operating very close to an uncooperative target, with the intention to get closer, berth with it, stabilize and eventually dock a third body onto it with the intention to have them both deorbit, requires high precision and high reliability as well as robustness of the system to non-nominal events. Challenges, that arise from handling a rocket body that is nearly 30 years old (the youngest SL-8 R/B of the chosen cluster was launched in 1986), exceed the fact, that the object was never designed to be deorbited: Pretending ADReS-A has already captured sight of the target and gets prepared to get closer, an imprecise knowledge about the tumbling rate requires sensors or cameras to record the actual rotation to calculate a safe berthing trajectory. On-board algorithm can help to determine the motion, starting with a more precise understanding of the tumbling and axis of rotation of the rocket body, and followed by the detection of the actual attitude orientation. A further algorithm will calculate a trajectory for docking and predetermine possible abort trajectories21. Knowledge about the targets movement and orientation is thus one requirement. Missing signal reflectors on the surface of the target, however, make the determination very difficult. Changing lighting conditions cannot guarantee for sufficient illumination of the target. For this reason, a Time of Flight Camera, which uses laser-impulses to develop a 3D-model, is implemented in the design. Even though the camera can work in darkness, illumination is recommended. By using this camera not only for the determination of the tumbling rate but also for evaluating the correct trajectory for approach, the implemented autonomy will have to deal with incomplete or corrupt data. One solution could be the application of previously defined dependency models of the subsystems and sensor data among themselves. The second requirement is thus the knowledge about the momentary relative distance of the two objects, a third requirement is the implementation of interacting dependencies. The next flaw is an absent predesigned point of contact for berthing. Reliable data about the specific SL-8 R/Bs is not available at this point of mission design and it is unclear, if this data still exists. With the thruster of a rocket body being designed to hold enormous heat over long periods of time, it can be

Page 7 of 12

66th International Astronautical Congress, Jerusalem, Israel. Copyright ©2015 by the International Astronautical Federation. All rights reserved.

assumed, that this is the most stable point of contact of the target. Research performed22 resulted in the KhimMash 11D49 engine as used thrusters for the SL-8 R/Bs. The heat shield of the upper stage or the nozzle itself shall be used as point of contact for the gripper, the detailed mechanism being out of the scope of this work, its nozzle is used for the connection and attachment of the De-Orbit Kit. The autonomy in this case needs to supervise the data transfer and make sure boundaries between the spacecraft and the target are kept. A following of the tumbling by the spacecraft during the fixture of the contact might be a solution, however, fast reaction will be required if the surface of the spacecraft gets too close to the rocket body itself to avoid contact on unintended areas with a collision as a result. An additional concern is possible leftover propellant in pressurized containers. A very carefully conducted capture operation will have to be implemented to reduce the possibility of explosion. It is impossible to predesign and implement all situation that might rise from this constellation on-board with the computational storage flown nowadays. Implementing autonomy that can react to unintended situations due to its self-awareness about its operational capabilities and margins as well as its work environment on improved processors could solve this problem. The short timeframe available in such situations will moreover most probably not allow for the ground control to intervene in time, and as mentioned before, the switch into safe mode so close to an uncontrolled object will most likely exacerbate the situation. A selfimplemented maneuver has to be performed to keep the spacecraft safe and operational, requiring re-planning capabilities by taking the changed situation into account. The listed considerations for the on-board autonomy of the ADR-mission need to be adapted and probably extended as the design process proceeds. III.III.

Concept

The concept of autonomy for the mission shall follow the three-tier (3T) architecture for system autonomy, as it has been proven to work for robots and is believed to have the capability to reach high-level autonomy. High level, in this context, refers to level E4 of the mission execution autonomy levels, defined by the European Cooperation for Space Standardization (ECSS23). Following their description, an autonomy level E4 executes goal-oriented mission operations onboard, and includes re-planning. The 3-layer architecture (also called three-tier architecture, short 3T) was developed in the mid-1980’s for robots, resulting from a slightly different approach24. 3T provides the possibility, to separate the three functional elements of sensing, planning and executing and have them work reactive. The executer in such way

IAC-15-D1.4.9

is based on stateless sensor-based algorithms, the sequencer is inhabited by algorithms that contain memory about the past and the planner (or deliberator) is made of algorithms that make predictions about the future. All these components run as separate computational processes, and exchange only specific data. The advantage of this approach is the physical separation of the layers, allowing a flexible and reusable approach with the layers being modified or added separately, even different software can be used. For ADReS-A, an additional characterization shall be used – the terms of A-Priori-Knowledge and Situational knowledge. The first of the two is modelled by the developer and thus represents the expert’s knowledge about the system and subsystems, generated during design time. The latter represents the actual situation and is created during runtime, based on the former but free to choose the best suitable solution for the appropriate situation. In the following, a description of the 3T-architecture for ADReS-A is given and how the two characterizations support the concept with Figure 5 displaying an overview. The first tier of the 3T architecture is the deliberative and thus time-consuming layer, responsible to process the planning. Based on the environmental status, such as the distance to the target and the spacecraft’s attitude, the task of the layer is to identify goals and concepts out of the pre-designed ones to proceed those goals with help of the pre-designed priority list. This list exists as there might be contradictory goals at the same time. For the defined goals and concepts, requirements, based on the spacecraft’s constraints, are taken into account to develop the task agenda. The agenda provides an order in which tasks have to be performed to achieve the goal(s). This leads to the second layer, in the 3T architecture called the sequencer, responsible to process the timing. Based on the current available resources, the tasks are now scheduled. Latest information on the constraints for the specific situation support the timing process. The over-all resources and constraints are provided by the developer (A-Priori-Knowledge), while the timing process is situational based, requiring information about the latest status of spacecraft and the work environment. With a schedule of the tasks to be performed developed, the third layer of executing can actually send commands to the different subsystems involved. As it is called controller or reactive layer in the 3T-architecture, and designed as such, the executing process needs to control the performance by examining the actual status and compare the outcome with the intended one. In case of a violation, a re-planning will be necessary. The depth of violation is an important aspect of the design of the autonomy. Margins need to implemented, that change situational based as some data is situation dependent with a variety of margin-possibilities.

Page 8 of 12

66th International Astronautical Congress, Jerusalem, Israel. Copyright ©2015 by the International Astronautical Federation. All rights reserved.

The requirements of the required autonomy for ADReS-A will be reflected in the pre-designed goals. Status from Environment Simulation

Re-Planning, based on New Status

Planning

Priority List Status-according goal & concept identification

Pre-Designed Goals & Concepts

Requirements

Constraints

Task Agenda

Task Options

Plan (Order) Timing Current Resources

Resources

Scheduler

Constraints Timer

Excecuting Task Execution

Procedures

Examination of Status

Command

Environment

Subsystems

Sensori-Motor Patterns

Control

Figure 5: Illustration of task distribution for ADReS-A, based on 3T-architecture (planning, timing, executing). Dark grey boxes represent the A-prioriKnowledge, light grey boxes the Situational Knowledge. Further implementations regarding the A-prioriknowledge need to be extracted from the spacecraft and mission design. The Situational Knowledge will be developed during runtime and thus representing the latest information about the spacecraft and the work environment. To have a stable strategy to create Situational Knowledge, an adequate concept needs to be chosen. One possible solution is presented in the following section. III.IV. Approach for ADReS-A Sensor Data Evaluation With the basic design of ADReS-A settled, sensors and actuators used are defined, as well as their position within in the system and the data they transfer to the processors. Due to different physical characters of the sensors used, the different phases of the mission, such as eclipse, illumination, trajectory from parking orbit to the holding point and the operational phase of berthing, require only parts of the available sensors. Holding point in this context refers to a stable point in a safe relative distance to the target. The sensors used during

IAC-15-D1.4.9

the operational phase of berthing are Star Tracker, CSSs, FSSs, IMUs, and the Time of Flight Camera (ToFC), their specific deployment during the phases was described earlier. A simulation to model the environment of ADReS-A to create self-awareness in terms of attitude and orientation of itself and the target and thus knowledge about the working environment is developed. The process on which the simulation is based evaluates the data send by the different sensors and will later on extract situations out of the changing values25. So far the simulation is a static model and needs further development to support the autonomy itself to work as required. Target Approach and Abort To berth with the target, the data of the ToFC is collected from a distance of about 11 m. Processed onboard with a specific algorithm12, the ToFC is able to track the motion of the rocket body and predict its future relative position and attitude. A further algorithm onboard21 based on the tumbling rate predictions, calculates the approach trajectory and transfers the signal to the thrusters. During the actual approach, every chosen step of the trajectory (e.g. every 5 s), the spacecraft analysis if the data send by the ToFC correlates with the predicted path. In case the path changes for any reason, a retreat command will be send. An adapted algorithm26 of the approach algorithm will have already calculated different paths (according to the chosen steps of 5 s) that can be taken if an abort is necessary. The calculations for the abort have to be performed before the approach starts and thus before the failure occurs as it will take too much time for calculations during the approach. Depending on the relative distance of the two objects, the abort will either command the spacecraft to retreat to the former holding point and wait from there for the ground station analysis, or retreat ‘in front’ of the target to a safe distance. As example for the latter, Figure 6 shows a screenshot of the simulation. The green path represents the already passed part of the trajectory, the red path is the pre-calculated and now followed abort. For calculation, the dynamics of the two objects need to be determined and predicted for the near future, both based on cylindrical geometries (cf. Figure 6). The docking maneuver is considered successful when the relative position and velocity of the docking points is zero. That allows for the robotic arm to grab the target and in such way create a fixed contact. ADReS-A shall be able to stay in this position for a predefined period of time, making a stable connection possible by giving the arm enough time to operate. The handling of stabilization of the new system would be the next step to succeed with active debris removal but is not part of this paper.

Page 9 of 12

66th International Astronautical Congress, Jerusalem, Israel. Copyright ©2015 by the International Astronautical Federation. All rights reserved.

Taking COSA as role model, the different sections (displayed in grey boxes in Figure 7) as well as the developer knowledge need more thoughts. However, with the requirements set and the spacecraft environment as well as the approach simulated, a promising basis is developed to implement autonomy for ADReS-A. IV. CONCLUSION AND NECESSARY FUTURE DEVELOPMENTS Figure 6: Simulated abortion trajectory26. The green line shows the intended approach, the red line follows an abort trajectory. Chaser and target are displayed simplified as cylinder. Future implementation of COSA So far, the simulation works static and more automated than autonomous. However, with the environment simulation of the spacecraft developed, the implementation of an autonomous architecture can be initiated. One possible solution that is promising to fulfil the previously mentioned requirements could be an architecture developed by the Institute of Flight Systems of the Universität der Bundeswehr München. The Cognitive System Architecture (COSA) has been successfully tested on unmanned aerial vehicles. UAVs have limited storage capabilities to process large amounts of data or perform on-board calculations to react to unpredicted situations, making them somewhat similar to spacecraft. The process of handling data is not exactly based on the 3T architecture, but can be transferred as COSA was initially build to support the human-robot interaction. Instead of the layer of planning, timing and executing, the layers are separated in skill-based, concept based and procedures based behaviors as displayed in Figure 7. The formerly mentioned A-Priori-Knowledge is displayed in dark blue color by pointing arrows, the Situational Knowledge, as it is created during runtime, is displayed in red color along the connecting arrows.

The paper addresses the necessity to implement high-level autonomy for active space debris removal missions. To set requirements for such autonomy, a mission concept and architecture have been developed with a preliminary design of the spacecraft itself and the de-orbiting devices introduced. Based on this mission, autonomy requirements are developed, open for further adjustment and extension. Finally, an initial approach for a more profound definition of the autonomy is given with COSA as promising architecture for further investigations. To actually realize a mission that addresses the active removal of space debris, legal or policy framework still has to be defined. The space community needs to work on this topic and come to a conclusion that enables active debris removal of objects that do not necessarily belong to the acting state. Moreover, funding concepts have to be developed as debris is a global issue involving all space faring nations and its customers. These concept can range from raising taxes to support ADR over insurance companies that perform ADR for their customers etc. Technical problems not solved involve a safe and reliable grabbing mechanism, the tracking of the target with the camera further away than mentioned in the text, and the stabilization part after a successful connection to the target. Additionally, more information regarding the target would be required to have a reliable basis for the calculations. Future developments within this project will concentrate on the autonomy concept for the ADReS-A mission. To implement COSA, actual data for the rules, constraints etc. need to be set. With the spacecraft and the de-orbiting device being designed on a preliminary basis, resource margins can be set as well as scheduling issues. The simulation allows now for goals to be designed and actions on how to reach those goals, including action alternatives, to be developed. Moreover, a priority list changing with the momentary situation has to be derived. Future development will therefore concentrate the design of these features with respect a minimal example.

Figure 7: Simplified concept of the COSA work structure

IAC-15-D1.4.9

Page 10 of 12

66th International Astronautical Congress, Jerusalem, Israel. Copyright ©2015 by the International Astronautical Federation. All rights reserved.

ACKNOWLEDGEMENTS This work is supported by Munich Aerospace and Helmholtz Association. The project ‘Sicherheit im Orbit’, which is the guiding theme for this work, is a cooperation between DLR and Universität der Bundeswehr München. Additionally, the authors would

like to thank Harvey Gomez for his support on the mission phases and the sensor’s utilization during such. Acknowledgements also go to the students of the Universität der Bundeswehr München that supported this work with their theses.27

1

IADC Space Debris Mitigation Guidelines. IADC-02-01, Revision 1; Sept 2007 Liou J.-C., Engineering and Technology Challenges for Active Debris Removal (2011), 4th European Conference for Aerospace Sciences (EUCASS) 3 Blasco A., Gil J., Graziano M., van der Linden B., Janssen B., Reynier P., Beck J., Herdrich G., Marynowski T. and Hatton J., Overview of the Results of ATV-1 Re-entry Observation Campaign (2011), International Astronautical Congress, Cape Town, South Africa, IAC-11-A6.2.10 4 Rank P., Mühlbauer Q., Naumann W. and Landzettel K., The DEOS Automation and Robotics Payload (2011), Symp. on Advanced Space Technologies in Robotics and Automation, ASTRA, the Netherlands 5 Nishida S.-I, Kawamoto S., Okawa Y., Terui F. and Kitamura S., Space debris removal system using a small satellite (2009), Acta Astronautica, 65 (1-2), pp. 95–102, doi: 10.1016/j.actaastro.2009.01.041 6 Debus T. J. and Dougherty S. P., Overview and Performance of the Front-End Robotics Enabling Near-Term Demonstration (FREND) Robotic Arm (2009), AIAA, 2009 (1870) 7 Sabelli E., Categorizing Admittance Control Parameters for the Ranger 8-DOF Tele-Operated Space Manipulator (2007), Space System Laboratory, University of Maryland 8 Scheper M., P²-ROTECT: Active De-orbiting of Large LEO Space Debris by OTV (2012). Task 5.2 9 Peters S., Förstner R. and Fiedler H., Mission Architecture for active Space Debris Removal using the Example of SL-8 Rocket Bodies (2015), Space Safety is No Accident, Springer International Publishing, pp. 23-28, doi: 10.1007/978-3-319-15982-9_3 10 Peters S., Fiedler H., Mai W. and Förstner R., Research Issues and Challenges in Autonomous Active Space Debris Removal (2013), International Astronautical Congress, Beijing, China, IAC-13-A6.5.3 11 Fehse W., Rendezvous with and Capture / Removal of non-cooperative Bodies in Orbit (2014), Journal of Space Safety Engineering 1(1), pp. 17-27 12 Gomez Martinez H. and Eisfeller B., Autonomous Determination of Spin Rate and Rotation Axis of Rocket Bodies based on Point Clouds (2016), SciTech San Diego, USA 13 GMAT. https://gmat.gsfc.nasa.gov/ 14 Bergler S., Untersuchung der Optimierungsmöglichkeiten für den Missionsablauf einer Multiple-Space-DebrisRemoval Mission (2015), ISTA-15-PA-11, Project Thesis at the Universität der Bundeswehr München 15 Udrea B. and Nyak Mikey, A Cooperative Multi-Satellite Mission for Controlled Active Debris Removal from Low Earth Orbit (2015), IEEE Aerospace Conference, Big Sky, Montana, USA 16 Outer Space Treaty (1967), Official Records of the General Assembly, Twenty-first Session, agenda items 30, 89 and 91, document A/6431 17 Liability Convention (1972), Official Records of the General Assembly, Twenty-sixth Session, Supplement No. 20 (A/8420) 18 Weeden B., Overview of the legal and policy challenges of orbital debris removal (2011), Space Policy 27, pp. 38-43, doi: 10.1016/j.spacepol.2010.12.019 19 Truszkowski W., Hallock H., Hinchey M., Karlin J., Rash J., Rouff C., and Sterritt R., (2009), Autonomous and Autonomic Systems: With Applications to NASA Intelligent Spacecraft Operations and Exploration Systems (2009), Springer-Verlag London, doi: 10.1007/978-1-84628-233-1 20 Wander A. and Förstner R., Innovative Fault Detection, Isolation and Recovery on-board Spacecraft: Study and Implementation using Cognitive Automation (2013), 2nd International Conference on Control and Fault Tolerant Systems, pp. 336-341, doi:10.1109/SysTol.2013.6693950 21 Michael J., Chudej K., Gerdts M. and Pannek J., Optimal Rendezvous Path Planning to an Uncontrolled Tumbling Target (2013), Automatic Control in Aerospace, Volume 19, Issue 1, pp. 347-352, doi:10.3182/201309025-DE-2040.00001 22 Pirzkall J. C., Development and Design of a De-Orbit Device (2015), ISTA-15-MA-05, Master Thesis at the Universität der Bundeswehr München 23 ECSS Executive Secretariat, (2008). Space Engineering: Space Segment Operability 2

IAC-15-D1.4.9

Page 11 of 12

66th International Astronautical Congress, Jerusalem, Israel. Copyright ©2015 by the International Astronautical Federation. All rights reserved.

24 Gat E., On Three-Layer Architectures (1998), AI-based mobile robots: Case studies of successful robot systems, pp. 195-210, Cambridge 25 Suttarp M., Entwicklung einer On-Board Architektur für ADReS-A (2015), ISTA-15-MA-04, Master Thesis at the Universität der Bundeswehr München 26 Schöpplein M., Erstellen einer Simulationsumgebung für das Docking eines Satelliten an ein Weltraumobjekt (2015), ISTA-15-SA-09, Seminar Thesis at the Universität der Bundeswehr München

IAC-15-D1.4.9

Page 12 of 12