FORMATION FLYING GUIDANCE FOR SPACE DEBRIS OBSERVATION, MANIPULATION AND CAPTURE

Transcription

1 ASTRONET II INTERNATIONAL FINAL CONFERENCE FORMATION FLYING GUIDANCE FOR SPACE DEBRIS OBSERVATION, MANIPULATION AND CAPTURE T. V. Peters (GMV) Property of GMV All rights reserved

2 OVERVIEW n Presentation loosely organized around mission phases, taking examples from different projects Engineering issues GNC aspects, with emphasis on guidance n Material from following projects: Detumbling: detumbling space debris after capture Patender: net capture tests COBRa: influencing debris (orbit and) attitude by plume impingement Android: demonstrate robotic and net capture of space debris edeorbit: de-orbit Envisat 2015/06/16 Page 2

3 INTRODUCTION n Introduction Space debris distribution Space debris dynamics Debris capture options n Mission phases for robotic capture Mid-range rendezvous Inspection from spiral orbit Attitude synchronization Capture and detumbling n Conclusion 2015/06/16 Page 3

4 FORMATION FLYING GUIDANCE FOR SPACE DEBRIS INTRODUCTION Property of GMV All rights reserved

5 DEBRIS CLASSIFICATION n Removal options Removal of tiny & small debris not practical Removal of large objects removes potential sources of fragments in case of collision Target selection is based on debris generating potential n Conclusion: remove large objects Type Characteristics Hazard Tiny Not tracked, <1 cm Shielding exists, damage to satellites may occur Small Not tracked, diameter 1 10 cm, 98% of lethal objects, ~ objects in LEO Medium Large Tracked, diameter >10 cm, <2 kg, 2% of lethal objects, ~ objects in LEO, > 99% of mass (incl. large objects) Tracked, >2 kg, <1% of lethal objects, > 99% of mass (incl. medium objects) Too small to track and avoid, too heavy to shield against Avoidance manoeuvres performed most often for this category Primary source of new small debris, 99% of collision area and mass 2015/06/16 Page 5

6 DEBRIS DISTRIBUTION n Debris population Total mass estimated at 6300 tons High concentration at inclination COSMOS 3M n SSO particularly important for and Earth observation and science SSO inclination-paired with inclination orbit Heightens collision probability Orbit planes may align, leading to head-on collisions during entire orbit instead of only at nodes 2015/06/16 Page 6

7 DEBRIS ATTITUDE DYNAMICS n No systematic survey for attitude n Several sources of data are available The spin rate of upper stages tends to slow down (1 & 2) Spin-ups have been observed, likely due to outgassing events (2) Envisat (3 & 4) Spin-up event occurred some time between april 2012 and november 2013 Rotation rate has been slowing down since then Rocket upper stages have generally been observed in a flat spin (i.e., non-axial) (5) Initial spin state of rocket bodies tends to be axial Therefore it is expected that a transition to a major axis spin occurs at some point due to energy damping 1. Boehnhardt, H., Koehnhke, H. and Seidel, A. 1989, The acceleration and the deceleration of the tumbling period of Rocket Intercosmos 11 during the first two years after launch, Astrophysics and Space Science, vol. 162, no. 2, p Williams, V., Meadows, A.J., 1978, Eddy current torques, air torques and the spin decay of cylindrical rocket bodies in orbit, Planetary and Space Science, vol. 26, 1978, p Bastida Virgili, B., Lemmens, S., Krag, H., 2014, Investigation on Envisat attitude motion, e.deorbit Workshop 4. Kucharski, D., Kirchner, G., Koidl, F., Fan, C., Carman, R., Moore, C., Feng, Q., 2014, Attitude and Spin Period of Space Debris Envisat Measured by Satellite Laser Ranging, IEEE Transactions on Geoscience and Remote Sensing, Vol. 52, Issue 12, pp , DOI /TGRS Santoni, F., Cordelli, E., Piergentili, F., 2013, "Determination of Disposed-Upper-Stage Attitude Motion by Ground-Based Optical Observations", Journal of Spacecraft and Rockets, Vol. 50, No. 3, pp , doi: /1.A /06/16 Page 7

8 DEBRIS ATTITUDE DYNAMICS Envisat n Rotation axis known n Characteristic decay time ~4.5 years COSMOS n Rotation around major axis n Characteristic decay time between 100 and 470 days, with a mean of 161 days and median of 129 days 2015/06/16 Page 8

9 CAPTURE METHODS Net Capture method Sensitivity to rotation rate Rotation related issues Structural issues low fast de-spin required to avoid tether wind-up around target may require measures to avoid breaking off pieces of target Grappling high synchronization required requires structural hard point Docking with nozzle high synchronization required requires non-steerable nozzle Tentacles high synchronization required Harpoon (Rigid) high synchronization required Harpoon (Non-rigid) low fast de-spin required to avoid tether wind-up around target Pushing sock air-bag high requires pre-capture de-spin Foam projection Ion-beam Shepherd Electrostatic tractor (only for GEOs) Magnetic tractor high low low low centrifugal forces may disrupt foam; requires pre-capture de-spin low sensitivity to spin rate; method may be used to control rotation low sensitivity to spin rate; method may be used to control rotation low sensitivity to spin rate; method may be used to control rotation may require structure not covered by MLI for firm grip requires strong structure for contact (e.g., honeycomb panels) and avoidance of propellant tanks requires strong structure for contact (e.g., honeycomb panels) and avoidance of propellant tanks may require measures to avoid breaking off pieces of target may require structure not covered by MLI for firm grip (i.e., MLI may tear off) none none none 2015/06/16 Page 9

10 ROBOTIC ARM CAPTURE n Precursor activities dealt with cooperative targets (attitude controlled, visual markers, grappling interfaces) ETVS-VII Orbital Express (DARPA program) n FREND (DARPA) performed on-ground demonstration of capture of uncooperative target debris n Other missions/concepts being investigated: DEOS (passive v.s. active chaser AOCS investigated) edeorbit (several robotic arm and tentacles configurations proposed, as well as net-based capture) ANDROID (double demonstration of robotic arm and net) ETS-VII (NASDA/JAXA) DEOS Orbital Express (DARPA) edeorbit concept from ESA CDF study 2015/06/16 Page 10

11 NET CAPTURE Net n Several studies to mature net capture technology (net design, net deployment strategy and mechanisms) Vbar m 1 I 1 ϑ 1 m2 I 2 2R 2 ϑ 2 Patender 2R 1 Scalable to debris mass and size Rbar Composed of a pyramidal, conical or plane net stowed in a canister with four masses (bullets) attached to net vertices Pneumatic or spring-driven ejection of bullets Tether connection after capture n Studies to investigate controllability AGADiR Controllability remains a difficult problem 2015/06/16 Page 11

12 FORMATION FLYING GUIDANCE FOR SPACE DEBRIS MISSION PHASES Property of GMV All rights reserved

13 ACTUATORS AND SENSORS Actuators n Depends on size of object n Separate thrusters for orbit raising / lowering & rendezvous Orbit raising / lowering: (2 x) 2 x 22 N (Android; stack mass 425 kg) 2 x 500 N (edeorbit; stack mass 9500 kg) Acceleration per thruster 0.05 m/s 2 Rendezvous (2 x) 8 x 1 N (Android; mass during rendezvous 298 kg) 28 x 22 N (edeorbit; mass during rendezvous 1700 kg) Sensors Acceleration per thruster 0.01 m/s 2 (edeorbit) / m/s2 (Android) Sensor Model Mass Power Range [kg] [W] [km] Performance Source Comments GPS Phoenix up to 2 m DLR WAC DVS " TSD 0.01 m short range; min range < 1m JENA LIDAR RVS long range OPTRONIK scanning 2015/06/16 Page 13

14 MID-RANGE RENDEZVOUS n Far-range rendezvous performed using TLE and GPS n TLE accuracy after 1 week of propagation: Radial: maximum error < 1.5 km => drift of 14 km per orbit Attitude Att G TGT TRACK Att. Cmd. Bias.25 km + 1σ of.1 km => drift of 3.3 km per orbit Cross-track: maximum error < 1.5 km Along-track: maximum error < 30 km n Is a detection & handover to WAC possible? NORAD TLE WAC DATA GPS DATA Translation Debris TLE Rel N WAC based Abs N GPS based - Rough estimate of relative state Relative state Trans G ΔV 1. Legendre P., Deguine B, Garmier R., Revelin B., 2006, Two Line Element Accuracy Assessment Based On A Mixture of Gaussian Laws, AIAA , AIAA/AAS Astrodynamics Specialist Conference and Exhibit, August 2006, Keystone, Colorado 2015/06/16 Page 14

15 MID-RANGE RENDEZVOUS Search phase n Detection limit 0.25 pixels Camera FOV 28 n Detection can occur between km distance Depending on size of target Uncertainty cone of Well within WAC FoV Handover to WAC is possible x range [m] number of pixels covered / object size comments max range WAC max range WAC max range WAC handover distance % of WAC FOV 1855 filled % of WAC FOV 3886 filled % of WAC FOV 6583 filled working distance target Detection range km 1.5 km 2ϕ ±5 km z chaser 2015/06/16 Page 15

16 MID-RANGE RENDEZVOUS n Phase 1 Difference in SMA larger than radial uncertainty in TLE ±2 km difference in SMA Vision-based navigation requires some radial motion for faster convergence Small target may lead to late detection n Phase 2 Terminal orbit may require specific relative geometry Lighting conditions Earth in background Ground contact Modulate drift to accommodate terminal conditions x x 4 km z 100 m z S1 δs.m.a. [km] S5 <1 km S4 S0.5 km S3b km ~ 4000 m S3a # of orbits # of orbits available for detection detection distance [km] S2b S2a 50 m S1 2 km 500 m 2015/06/16 Page 16

17 MID-RANGE RENDEZVOUS Full guidance Guidance as implemented mode manager commands Guidance function Guidance function target orbit Mean orbit Kepler orbit target orbit chaser LVLH state Mean orbit Kepler orbit Plan generation Situation assessment Plan database Guidance expert function chaser LVLH state time Plan generation Table of times, reference states & ΔV s time Table of times, reference states & ΔV s correction time check Reference trajectory reference trajectory correction time check Reference trajectory reference trajectory ΔV computation ΔV ΔV computation ΔV 2015/06/16 Page 17

18 MID-RANGE RENDEZVOUS n n n n Initial errors are quite large After first 5 hours (3 orbits) errors decrease Decrease occurs when chaser enters drift orbit Large errors in position and velocity occur at large distances Errors in position and velocity show a slight increase over time due to the fact that a unperturbed Keplerian propagator is used to propagate relative trajectories Keplerian orbit is initialized at start of simulation starts diverging from true orbit over time Causes of errors are known, and could be improved. J2-based relative propagator could be used to improve the reference trajectory Guidance could be made to operate on linearized differential orbital elements Suffer less from linearization errors Guidance could periodically update its plan Re-initializing Keplerian orbit used to generate reference trajectory Reference orbit will be closer to true orbit and reference trajectory will be closer to truth x [m] Position error time [h] v x [m/s] ΔV [m/s] time [h] 0 Velocity error time [h] 2015/06/16 Page 18

19 INSPECTION FROM SPIRAL ORBIT n Camera in target pointing when constraints are met n Several constraints shall be taken into account: Eclipse times (no operation) Sun exclusion angle (50deg) Earth in the field of view (IP problems) Illumination conditions, angle Sun Picard Mango below 90 deg (IP problems) n When all constraints taken into account only about 25% of the orbit is useful, +ZY in LVLH Z [m] Y [m] D LVLH Relative trajectory X [m] 3D LVLH Relative trajectory X [m] Z [m] 3D LVLH Relative trajectory Y [m] /06/16 Page 19

20 INSPECTION FROM SPIRAL ORBIT n Effect of perturbations (SRP and Drag) lead to nonconstant drift rate Needs to be taken into account in manoeuvre definition for spiral orbit insertion if in drift-free orbit Leads to more correction manoeuvres Note: size of spiral orbit fairly small; 10 m vs m for COSMOS - Envisat n Possible input to spacecraft design to ensure small differences in ballistic coefficient Y [m] LVLH relative trajectory X [m] LVLH relative trajectory 5 LVLH relative trajectory Z [m] X [m] LVLH relative trajectory Z [m] Z [m] Y [m] Y [m] X [m] 2015/06/16 Page 20

21 SYNCHRONIZATION PHASE Elements n Target attitude propagation n Reference frame transformations n Quaternion spline curve for fly-around to limit accelerations n Straight-line approach using rampconstant-ramp velocity profile A S3 S2 Guidance plan consists of: X LVLH 1. Perform station keeping on Vbar 2. Transfer to target ω-vector (h-vector is an alternative) 3. Station-keeping at target ω-vector 4. Rotate to target co-rotating 5. Perform transfer closer to target 6. Station-keeping at w-vector 7. Transfer to target body fixed frame position 8. Station-keeping in body fixed frame position 9. Transfer closer to target Z LVLH 1. Myoung-Jun Kim, Myung-Soo Kim, and Sung Yong Shin A general construction scheme for unit quaternion curves with simple high order derivatives. In Proceedings of the 22nd annual conference on Computer graphics and interactive techniques (SIGGRAPH '95), Susan G. Mair and Robert Cook (Eds.). ACM, New York, NY, USA, DOI= / S4 S5 S1 2015/06/16 Page 21

22 SYNCHRONIZATION PHASE n Synchronization simulated in simplified simulator Propagation models Trajectory propagator contains J2 perturbation Attitude propagation propagates torque-free tumbling motion (no perturbations) Sensors and actuators 1 N thrusters with thruster management function Perfect sensors GNC Synchronization guidance LQR controller Perfect navigation 2015/06/16 Page 22

23 SYNCHRONIZATION PHASE Camera view LVLH view ω = 0.5 /s 2015/06/16 Page 23

24 SYNCHRONIZATION n Guidance trajectory is precisely followed Centimetre level accuracy in position Millimetre per second level accuracy in position Pointing error smaller than 0.1 Better accuracy possible with more aggressive controller n But, no navigation is included 2015/06/16 Page 24

25 SYNCHRONIZATION n ΔV required for synchronization guidance ΔV is red true ΔV is black n Four phases can be distinguished transfer to the angular velocity vector fairly expensive; sharp increase in ΔV right at start station-keeping at the angular velocity vector comparatively cheap; gradual increase in ΔV transfer to the body fixed frame station-keeping in body fixed frame Station-keeping in target reference frame is cheaper than transfer But considerably more expensive than station-keeping at angular velocity vector Especially considering that station-keeping at angular velocity vector is performed at 10 m, while station-keeping in target body frame is performed at between 2 and 5 m ΔV [m/s] time [min] 2015/06/16 Page 25

26 CAPTURE AND DETUMBLING n Capture and detumbling currently under investigation Inverse kinematics dependent on design of arm; full implementation provides low added value compared to cost Some simplification (e.g. convenient arm/joints configuration and allow small joint angles w.r.t. rigid) Contact model between robot hand and target approximated by translational and rotational spring damper Kc Dc system no gripper is attached to end point of manipulator, contact occurs just between two points state of spring damper could supply a metric on what is happening at contact 1. Dimitrov, D. N., Kazuya Y., 2004, "Momentum distribution in a space manipulator for facilitating the post-impact control," Intelligent Robots and Systems, 2004.(IROS 2004). Proceedings IEEE/RSJ International Conference on, vol. 4, pp IEEE, /06/16 Page 26

27 FORMATION FLYING GUIDANCE FOR SPACE DEBRIS CONCLUSION Property of GMV All rights reserved

28 CONCLUSION n Brief outline of mission phases for a debris capture mission has been presented Overview of results of several related projects n Envisat most likely candidate for a debris removal mission Exceptional for high rotation rate Special measures to be taken for attitude synchronization n Smaller ADR demonstration mission with a smaller target should be implemented 2015/06/16 Page 28

29 Thank you T. V. Peters Property of GMV All rights reserved