Twitter Facebook LinkedIn YouTube

Research Topic:

My research consist of the analysis and design of technologies and techniques for robotic active debris removal (ADR). More in detail, my research focuses on: (a) taxonomy of space debris based on most suitable ADR techniques, (b) analysis and design of a new robotic concept for ADR, (c) development and implementation of a robotic control to autonomously capture a tumbling target. There has been a great deal of fundamental research in almost all areas of space robotics and although capturing a tumbling target in orbit is a well known problem, to date, no spacecraft has ever performed an autonomous capture of a non-cooperative vehicle. Autonomy is requested in particular in the nal phases of the capture due to the limited reaction time available to face anomalies and/or communication problems that might occur. Moreover, it is very challenging to study a robotic ADR mission as a whole, due to different characteristics of mission phases, which is why most of the researchers tend to study those phases individually. Thus, I think that despite the great amount of existing literature there is still a lot of work to be done in order to bring to reality routine robotic ADR missions.

Main Results:

My main results of the research achieved to date consist of: (a) development of a new concept of robotic spacecraft for future ADR missions [1, 2], (b) outline of a GNC architecture for such a robotic spacecraft [3, 2], (c) formulation of the control techniques of the robotic payload to successfully capture a non-cooperative tumbling target. More in detail about the achieved results can be found in what follows. At rst an analysis of the current space debris environment was made (based on the results in the literature) which has led to the identification of SL-8 (i. e. Kosmos 3M) rocket bodies (R/Bs) as the most suitable targets for an ADR removal and stabilization of the current space debris population [3]. With that in mind a reference mission was outlined as a one year mission aiming at capturing five, intact, pre-selected R/Bs and de-orbiting them with either a hybrid propulsion modules (HPM) or a solid propellant...

Table 1: Capturing device trade-off

... kit (SPK). The chaser S/C is to be launched directly into the orbital plane of the rst target. The selected launcher of choice is an Ariane 5. The developed concept resulted in two configurations of the chaser S/Cs (see Fig. 1-5), both featuring a robotic manipulator for the inspection of the target and placement of a de-orbiting kit into the main nozzle of the R/B. The difference between them is in the device used to capture and rigidly connect to the target vehicle. In fact, the configuration in Fig. 1 and 2 has a semi-rigid clamping mechanism while the one in Fig. 4 and 5 uses a robotic arm for the same purpose. These configurations are a result of a trade-o study summarized in Tab. 1, where the number 5 indicates a maximum mark, while number 1 indicates the lowest one. The resemblance of the S/C to the European Automated Transfer Vehicle (ATV) is intentional in order to reduce the overall cost of the mission and raise the technology readiness level (TRL) of the spacecraft. The general characteristics of the base S/C are summarized in the Tab. 2.

The advantage of the semi-rigid clamp consist in: being able to capture the target despite surface irregularities, it does not require particular features to be grasped on the target body, it can be folded during the launch, its control is fairly simple and it can maintain the target rigidly connected to the chaser even in the event of the total power cut-o of the S/C. However, it requires relative motion synchronization prior to capture and a relatively strict AOCS uncertainty box in order to securely perform the capture. Moreover, it requires a non-trivial manufacturing and it will result in a non-selective application of the pressure during the capture. With the robotic arm the uncertainty box is more relaxed and with a proper control of the robotic arm the synchronization of the relative motion is not required. Nonetheless, the control of the robotic arm is more complicated in space (due to the dynamical coupling between the base and the manipulator) and is more delicate.

Table 2: Spacecraft specifications

In both cases the GNC system of the spacecraft is assumed to be switched off during the capture maneuver. The design of the semi-rigid, deployable clamp was based on the prototype of a finger of a humanoid robot, illustrated in Figure 3, made of a thermoset plastic insets, used as interphalangeal articulations, a set of rubber and metallic phalangeal bones and a set of strings used as tendons. Its design permits it to be stowable and light while at the same time being able to embrace the target before touching it. Its ability to dissipate any bouncing energy should be feasible due to its semi-rigid nature. The total estimated mass of the device including its supporting structure is around 240 kg. The manipulators on the other hand are envisioned to be an adapted version of the 7 degrees of freedom (DOF), torque controlled, light weight manipulator that will be used in the upcoming DEOS mission of the German Aerospace Agency (DRL) with custommade end effectors (EE) for the manipulation of de-orbiting kits and capture of target R/Bs.

In order to raise the TRL of the technologies needed for the previously mentioned mission, an additional concept of the spacecraft (visible in Fig. 6 and 7) was developed, within the SCARAB working group of the Stardust project, consisting of a demonstration spacecraft to actively de-tumble, capture and de-orbit its own launcher (an Ariane 5 upper stage). The main characteristics of the chaser spacecraft are: a deployable semirigid clamping mechanism, a de-tumbling device, a de-orbiting kit and a robotic arm.
The overall specifications of the S/C are listed in Table 3. More information about the mentioned spacecraft can be found in [2].
The guidance, navigation and control system has to: a) process the information coming form sensors, b) plan the execution of appropriate maneuvers and c) perform them.
Based on the following, the GNC architecture (visible in Fig. 8) has been defined to support future robotic ADR missions, like those described previously, consisting of several software modules, each responsible for a particular function within the GNC system. More in detail, most suitable algorithms for populating those modules were identified...

Figure 1: S/C concept for ADR of Kosmos 3M R/Bs: clamp configuration

Table 3: Agora chaser spacecraft specifications [2]

Figure 2: S/C concept for ADR of Kosmos 3M R/Bs: clamp conguration

Figure 3: Scaled prototype of on of the ngers of the grabbing device [1]

Figure 4: S/C concept for ADR of Kosmos 3M R/Bs: manipulator conguration

Figure 5: S/C concept for ADR of Kosmos 3M R/Bs: manipulator conguratio

Figure 6: S/C concept for ADR of Ariane 5 R/Bs [1, 2]

Figure 7: S/C concept for ADR of Ariane 5 R/Bs [1, 2]

and characterized , having in mind the most critical phase of any ADR mission which is the close range rendezvous. The exclusion of the robotics module from the GNC architecture is intentional due to the necessity to deactivate the GNC during the capture
and improve the overall performance of the system due to the complex dynamics at hand. Two algorithms were identified for each module, having in mind that the first one should be used only as a baseline (given its space heritage) while the second one should represent the evolution of the baseline [3]. For more details about the selected algorithms please refer to [3].

The tasks of the robotics module are essentially to:
1. control the capture of a tumbling target, by means of a robotic manipulator
2. stabilize the compound (i. e. chaser plus target), while limiting the transfer of the angular momentum from the target to the chaser

The developed concept of the robotic control architecture, illustrated in Figure 9, is divided into two modules: an onboard (i. e. on-line) and on ground (i. e. o-line). The latter consists of target motion simulation and prediction module along with a motion planner based on learning algorithms. The former uses the calculated o-line solution as an initial guess for the trajectory generation and control of the robotic arm in real time. The reason behind this division lays in the computational requirements of the motion planner based on machine learning, that can not be performed in a reasonable time with the computational power of nowadays onboard computers [3]. Thus, currently a local trajectory generation method has been formalized based on the backward integration of the Bias Momentum Control (BMC) approach developed originally by Dimitrov D. N. et al.

Figure 8: Autonomous onboard rendezvous and capture control system architecture [3]

Figure 9: Concept of robotic control architecture [3]

The novelty of the method stands in considering a non-zero initial angular momentum of the base spacecraft, deactivated AOCS of the spacecraft during the capture, synchronization time needed to successfully capture a tumbling target and the analytical approach that makes this method suitable for an onboard implementation. Its implementation and validation are future steps to be preformed.

[1] M. Jankovic, K. Kumar, N. Ortiz Gómez, J. Romero, F. Kirchner, F. Topputo, S. Walker, and V. M., Spacecraft Concept for Active De-Tumbling and Robotic Capture of Ariane Rocket Bodies-Poster, in ASTRA 2015 Proceedings, (Noordwijk, the Netherlands), ESA, European Space Agency, May 2015.

[2] M. Jankovic, K. Kumar, N. Ortiz Gómez, J. Romero, F. Kirchner, F. Topputo, S. Walker, and M. Vasile, Robotic System for Active Debris Removal: Requirements, State-of-the-Art and Concept Architecture of the Rendezvous and Capture (RVC) Control System, in 5th CEAS Air & Space Conference Proceedings, (Delft, the Netherlands), CEAS, September 2015.

[3] M. Jankovic, J. Paul, and F. Kirchner, GNC architecture for autonomous robotic capture of a non-cooperative target: preliminary concept design, Advances in Space Research, May 2015.

[4] M. Jankovic, J. Paul and M. Vasile, "Stardust Initial Training Network (poster)", in 3rd European Workshop on Space Debris Modelling and Remediation, CNES Headquarters, 2 place Maurice Quentin, Paris, France, 2014, p. single page.

[5] M. Jankovic, K. Kumar, N. Ortiz Gomez, J. Romero Martin, F. Kirchner, F. Topputo, S. Walker and M. Vasile, "Spacecraft Concept for Active De-Tumbling and Robotic Capture of Ariane Rocket Bodies (poster)", in Advanced Space Technologies in Robotics and Automation-ASTRA 2015, ESA - ESTEC, Noordwijk, The Netherlands, 2015, p. single page.

1. Personal webpage at DFKI:
2. LinkedIn prole webpage:
3. ResearchGate webpage:

Funded by