Content uploaded by Mathieu Deremetz
Author content
All content in this area was uploaded by Mathieu Deremetz on Oct 27, 2020
Content may be subject to copyright.
Page 1 of 10
MOSAR-WM: A relocatable robotic arm demonstrator for future on-orbit applications
Mathieu Deremetza*, Pierre Letiera, Gerhard Grunwaldb, Máximo A. Roab, Bernhard Brunnerb, Benoit
Lietaera, Michel Ilzkovitza
a Space Applications Services NV/SA, Leuvensesteenweg 325, 1932 Sint-Stevens-Woluwe (Brussels Area), Belgium,
firstname.lastname@spaceapplications.com
b Institute of Robotics and Mechatronics, German Aerospace Center (DLR), 82234 Wessling, Germany,
firstname.lastname@dlr.de
* Corresponding Author
Abstract
In the past few years, the raise of space robotics yielded novel potential applications. The utilization of more
advanced and capable robotic manipulators opens a whole new horizon of possibilities for future space missions,
ranging from On-Orbit Servicing (OOS) of existing satellites (for refuelling, Orbital replacement unit (ORU) or de-
orbiting) to On-Orbit Assembly (OOA) and reconfiguration of modular spacecraft. This paper deals with the design
and primary Manufacturing, Assembly, Integration and Testing (MAIT) activities of a novel robotic manipulator
demonstrator for such on-orbit applications. MOSAR-WM is a 7 degree of freedom (DOF) manipulator, 1.6-meter
long, symmetrical and relocatable (aka. “walking” capable). Its overall structure is human-like with asymmetric
joints. Manipulator joints are hollow-shaft for internal cable routing, and include cutting-edge space-compatible
technologies. Each joint embeds a torque sensor in addition to position sensors (incremental and absolute encoders).
The kinematic architecture of MOSAR-WM offers a wide end effector workspace, and its stiff structure guarantees a
high accuracy and repeatability while allowing compactness for launching and storing purposes. Each extremity of
MOSAR-WM is equipped with a HOTDOCK standard interface that allows for mechanical connection, powering
and controlling the arm. Manipulator avionics consists in seven joint controllers (one per joint) and an embedded
computer called Walking manipulator controller (WMC) running a real time operating system. The WMC receives
high-level commands from the external computing unit through the connected HOTDOCK interface. It also
calculates the dynamic model of the robot to provide proper feed-forward terms for the joint control. Depending on
the desired behaviour, the gains of the joint control loop are adaptive for optimal performance in position control. In
addition, a Cartesian impedance control is implemented to allow for compliant operations. The joint controllers are
daisy-chained through EtherCAT, while the control of each HOTDOCK is performed through a CAN bus managed
by the internal WMC. MOSAR-WM is developed in the context of the European Commission’s Space Robotic
H2020 MOSAR project. It aims to validate the developed technologies at Technology Readiness level (TRL) 4 in a
space representative scenario.
Keywords: Space robotics, Relocatable Robot, Mechanism Design of Manipulators, Motion Control of Manipulators
Acronyms/Abbreviations
Walking Manipulator (WM), Extra-Vehicular
Activity (EVA), Payload Deployment and Retrieval
System (PDRS), Shuttle Remote Manipulator System
(SRMS), Orbiter Boom Sensor System (OBSS), Mobile
Servicing System (MSS), Mobile Remote Servicer Base
System (MBS), Space Station Remote Manipulator
System (SSRMS), Special Purpose Dexterous
Manipulator (SPDM), Japanese Experiment Module
Remote Manipulator System (JEM-RMS), Robot
Component Verification on ISS (ROKVISS), European
Robotic Arm (ERA), Standard interconnect (SI),
Walking Manipulator Controller (WMC). Telemetry
(TM), Telecommand (TC), Tool Centre Point (TCP).
Fig. 1. Artist representation of the MOSAR project
Page 2 of 10
1. Introduction
The growing interest in on-orbit assembly coupled
with the raise of advanced robotic systems is stimulating
the development of new paradigms for future space
missions [1]. In this context, autonomous or remote-
controlled robotic manipulators are becoming widely
used for diverse space applications, ranging from on-
orbit servicing of existing satellites (for refuelling, ORU
or de-orbiting) to on-orbit assembly and reconfiguration
of modular spacecrafts [2]. Robot manipulators are a
relevant technology for highly dexterous and accurate
operations in extreme environments such as outer space.
They are also able to perform position and force control
tasks with compliance, capable of supporting
cooperative tasks (EVA) or autonomous ones.
Robotic manipulators are present in space since
decades. Well known on-orbit applications are the
different robotic systems embedded in the space shuttle
and attached to the International Space Station, namely
PDRS (composed of SRMS and OBSS), MSS
(composed of MBS, SSRMS, OBSS, SPDM), JEM-
RMS, Strela, ROKVISS and later on ERA. They have
been progressively installed year after year to enhance
the maintenance abilities as well as assessing new
technologies on-orbit. ETS-VII is also notable as the
first unmanned satellite (experimental) equipped with a
robotic arm for performing autonomous rendezvous
docking operations.
Other robotic manipulators have been developed for
targeting future on-orbit missions including among
other: DEXARM [3] a robot arm comparable to human
arm for human equivalent intervention and applied in
EUROBOT a three-arm robot for supporting an ISS
crewmember during EVAs; CAESAR [4] a robot
system for on-orbit services, or even DSXR a self-
deployable and relocatable manipulator for assembling,
berthing and inspecting the lunar orbital platform-
gateway (LOP-G).
Aforementioned robotic systems are mainly
targeting conventional or dedicated missions; however,
new paradigms involving modular spacecraft and
standardisation of components are also emerging,
requiring the development of new advanced robotic
systems and related technologies.
H2020 EU funded project MOSAR [5] is one of
these new developments. MOSAR aims at designing
modular spacecraft and related key technologies to
enable on-orbit assembly and reconfiguration, as
illustrated in Fig. 1. It involves two types of spacecraft:
a modular client satellite equipped with replaceable
modules, and a servicer, bringing replacement modules.
The connection between each subsystem (spacecraft or
modules) is performed through standard interconnects
(SI), and a repositionable walking robot (MOSAR-WM)
allows performing autonomous assembly and
reconfiguration tasks.
The structure of this paper is as follows: Sec. 2
provides an overview of the MOSAR-WM system. Sec.
3 to 6 detail the design aspects related to MOSAR-
WM’s mechatronic subsystems, avionics, software and
control. Sec. 7 presents the initial results of the
manufacturing, assembly, integration and testing phase
and Sec. 8 provides a conclusion on the work achieved
so far and presents the perspectives on future activities.
2. System Overview
The two basic operations considered for the walking
manipulator (WM) are (1) re-localization to a new
attachment point on the satellite, and (2) manipulation
of a spacecraft module using the SI attached to the arm
(Fig. 2 & 3).
Fig. 2. MOSAR-WM re-localization operation.
Fig. 3. MOSAR-WM manipulation operation.
Page 3 of 10
The walking manipulator comprises the following
elements:
Structure (Limbs): The walking manipulator is
composed by eight structural elements. The
overall configuration of the manipulator is
based on a human-like arm with asymmetric
joints (See Fig. 4 & 5). The arm is 1.6-meter
long and weighs approximately 30kg.
Fig. 4. Overview of the MOSAR-WM.
Motorization (Robotic joints): The walking
manipulator is equipped with seven revolute
joints according to the following symmetric
configuration R┴R┴R┴R┴R┴R┴R, where R
indicates a revolute joint and ┴ the
orthogonality between two successive joint
axes (see Fig. 6). Motors have been sized for
lifting a 10kg payload at 1g across the entire
workspace of the manipulator.
Fig. 5. Limb description for the MOSAR-WM.
End effectors (HOTDOCKs): Each extremity
of the walking manipulator is equipped with a
Standard Interconnect (SI), namely
HOTDOCK as illustrated in Fig. 5. These
interfaces allow the robot to relocate, since the
arm can be attached to a supporting structure
on both sides.
Fig. 6. Axis definition for the MOSAR-WM.
Avionics: In order to perform motions, control
its robotic joints and receive/forward power
and data through its structure, the walking
manipulator is equipped with independent
avionics as illustrated in Fig. 7. Each motor is
driven by a dedicated joint controller (driver)
while the control of the WM displacement at
arm level is performed by an embedded
computer called Walking manipulator
controller (WMC) running a real time
operating system.
Fig. 7. Integration of the avionics inside MOSAR-WM.
All the motions of the WM are mainly based on two
types of elementary actions:
Transfer motion: it is the action to move the
WM free end-effector from its initial position
to a target position in joint space. It is typically
used for coarse collision-free motions of the
WM during the different operations. The first
step consists of configuring the WMC in
Position control mode, which is then replicated
to the configuration of the joint drivers. Then
Page 4 of 10
the joint transfer trajectory is sent by RMAP
command on the SpaceWire bus to the WMC,
that interpolates the sequence of points and
define the successive position set-points for the
joint drivers. Based on the feedback of the joint
sensors Telemetry (TM), the WMC confirms to
the spacecraft On-Board Computer (OBC)
when it reaches the goal position.
Approach motion: it is the action to perform
the final approach of the WM end-effector to
reach the position ready for the connection of
the end-effector SI to the payload or satellite
bus (also equipped with a SI). A similar motion
is also applicable when a spacecraft module is
manipulated by the WM to align with another
module and/or the satellite bus. The motion is
based on a Cartesian trajectory in order to
ensure as much as possible a suitable approach
angle. In order to mitigate potential initial
alignment errors, an impedance control scheme
is applied to enable the guidance using the
form fits of the SI, by estimating the contact
forces using the joint torque sensor. The first
step consists of configuring the WMC in
impedance control mode, which is then
replicated in the configuration of the joint
drivers. Then the Cartesian trajectory is sent by
RMAP command to the WMC, which
performs kinematic conversions and dynamic
computations to derive the required set-points
for the joint drivers. Based on the feedback of
the joint sensors TM (position, torque), the
WMC iterates and updates the set-point until
the SI TM confirms their proper alignment.
The SI are then ready for latching/connection.
This information is confirmed to the spacecraft
OBC through the RMAP TM (either from the
WM end-effector SI or the satellite module SI),
which then sends the command to stop the
process. In order to ensure proper contacts of
the surfaces, it is also possible to have a local
validation of the alignment in the WM itself by
using the estimated contact force.
3. Mechatronic Subsystems
This section describes the main mechanical
subsystems developed and embedded in MOSAR-WM:
the standard interfaces and the motor joints.
4.1 MOSAR-WM end-effectors
The two tips of the walking manipulator are equipped
with a SI, namely HOTDOCK, depicted in Fig. 8.
HOTDOCK is an androgynous standard robotic mating
interface, developed by Space Applications Services
NV (Belgium), supporting mechanical, data, power and
thermal transfer. It is primarily designed to allow on-
orbit modular assembly and reconfiguration of
spacecraft elements. HOTDOCK simplifies the
replacement of failed modules and allows for payloads
swapping. At the same time, HOTDOCK provides
chainable data interfaces for multiple module
configurations.
Fig. 8. HOTDOCK standard robotic mating interface.
HOTDOCK features the following coupling elements:
A mechanical interface that provides:
- alignment;
- coupling/connection;
- mechanical load transfer capability.
It is composed of fixed elements (main body
structure and the form fit geometry) and a
movable locking ring to allow connection with
another device. Furthermore, the geometry of
HOTDOCK makes it androgynous and 90-
degree symmetrical.
A central functional interface comprising a
contact plate with spring-loaded pins and pads
that enables:
- transfer of electrical power;
- transmission of data through CAN or
SpaceWire protocol;
HOTDOCK embeds its own controller for local
management (actuator, sensor, TM/TC communication)
and rear connectors harnessing giving access to the
power/data interface pins and the internal
controller/powering of the device.
Page 5 of 10
Key characteristics of HOTDOCK are summarized
below:
TRL: 4
mass: 1.5kg
approach admissible misalignment: +/-15mm
and 10 degrees
maximum approach angle: 60 degrees
coupling admissible misalignment: 2mm
coupling time when positioned: 20-30sec
Maximum voltage: 120V
Maximum current: 20A
Maximum data rate: 200Mbit/s
4.2 MOSAR-WM joints
Each joint of the walking manipulator is equipped with
a rotary hollow shaft actuator for enabling cable routing
inside the arm.
The design of the joints is based on the following
components:
Motor:
Frameless Brushless DC (BLDC) motor. This
type of motor offers the following features:
design flexibility, hollow shaft, high torque
density and compactness.
Reducer:
Strain wave gear to transmit the required output
torque. This type of reducer offers the
following features: backlash-free, extremely
high positioning accuracy, excellent
repeatability, high torque capacity, high
efficiency, high ratio in a compact and simple
design.
Sensors:
Magnetic Incremental encoder to provide an
accurate motor shaft position. Used for the
commutation and input speed control.
Magnetic Absolute encoder to provide an
accurate output shaft position.
Torque sensor to provide an accurate measure
of the output shaft torque. Used for the
compliance control of the arm.
Temperature sensor to monitor the temperature
of the motor.
Safety:
Electromagnetic brake mounted upstream the
reducer to ensure a maximal braking torque.
Electrical and mechanical end-stops mounted
on the structure to respectively notify end
positions of the joints and block mechanically
the joints if needed.
Guiding:
Precision Bearings to guide accurately the
motor and output shafts, and also carry the
loads with a high stiffness.
Two actuator sizes, depicted in Fig. 9, have been
designed to fulfil the requirements linked to the Earth
demonstrator operations.
Fig. 9. Joint drive units – elbow/wrist (left) and
shoulder (right).
Key features of the elbow/wrist and shoulder joints are
respectively:
Nominal Torque: 170Nm, 300Nm
Maximum output speed: 3rpm
Accuracy: +/-0.025deg/90arcsec
Repeatability: <0.001deg
Weight: 2.4kg, 3.6kg
4. Avionics & Electrical Architecture
Fig. 10 describes the detailed WM avionics and
electrical architecture. The WM Controller (WMC) is
the central avionics component. It manages the internal
control of the arm, of the two HOTDOCK at the WM
tips, and the communication with the satellite OBC. The
WMC uses an Intel NUC board, running Ubuntu 18.04
LTS 64-bit. This solution, equivalent to the spacecraft
OBCs, offers the required compactness while providing
a standard platform to integrate the different software
applications.
The WMC interfaces the three main data buses
running along the robotic arm:
SpaceWire bus for “high level” non-
deterministic TM/TC communication with the
servicer OBC through the RMAP protocol.
This relies on a Star-Dundee SpaceWire Brick
Mk3, connected by USB to the WMC. The
Page 6 of 10
Brick also ensures the transmission of the
SpaceWire signal between the two HOTDOCK
extremities.
CAN bus for local control of the HOTDOCK
interfaces. This is done with a USB-CAN
interface.
EtherCAT bus for low-level deterministic
communication with the joint drivers. This is
done by the implementation of an EtherCAT
Master application on the WMC and
replacement of the Ethernet drivers by
EtherCAT driver, to allow the connection
directly though the WMC Ethernet port.
Each joint of the WM is equipped with a local
controller for the closed loop position/current control of
the actuator and the measurements of the joint sensors.
All joint controllers are interfaced through the
EtherCAT bus, managed by the WMC, which ensures
real-time exchange of information required for the
control algorithm of the arm at 2 kHz.
The robotic arm is powered by the servicer 48V
power bus, through the power interface connector of
HOTDOCK. The 48V bus is directly interfaced with the
joint drivers that will manage the power interfaces to the
motors and sensors of the joints. Local DC/DC
convertors provide the required 24V to power the two
HOTDOCKs and 19V for the WMC. The arm is able to
control the power transfer, passing through the
interfaces, thanks to power relays embedded in
HOTDOCK. This allows to power-on a module
connected at the end-effector and communicate with it
through SpaceWire.
5. Software Architecture
In the MOSAR application, the WM is connected to
the spacecraft OBC through a chain of SpaceWire links.
This might introduce unpredictable latency affecting the
impedance control of the arm. Impedance control is
needed to press the standard interfaces together,
mitigating the possible misalignments to enable a
successful locking of the mechanism. This is handled
locally by the WM Controller, which reads the torque
measurements and commands the joints at a very high
rate
The software running on the WMC is depicted in Fig.
11. It consists of three branches, each of them
associated with a specific data bus of the robot:
The TMC Engine is the software module in
charge of the SpaceWire bus interface and
performs the conversion of the high-level
commands received from the spacecraft OBC
through RMAP in the appropriate low-level
commands required for the control of the WM
arm and the HOTDOCK end-effectors.
Fig. 10. Avionics and electrical architecture
Page 7 of 10
Fig. 11. Software architecture
The Robot Monitor and Control software
module is in charge of the management and
control of the robotic arm. It is interfaced with
the EtherCAT handler, which is based on the
open source EtherLAB Master library. The
EtherCAT handler runs in real-time, to ensure
good performances of the control algorithms.
Specific attention is given to limit the jitters
and delays in the communication (e.g. real-
time patch, CPU isolation….).
The HOTDOCK Monitor and Control software
module is in charge of the management and
control of the two HOTDOCK end-effectors of
the arm that are interfaced to it through the
CAN bus.
6. Robot Control
The control of the MOSAR-WM is based on
impedance controllers, which allow a modulation of the
mechanical impedance at the joint or arm level while
interacting with the environment. In this way, the
system successfully completes interaction tasks such as
docking or grasping objects, while making it very robust
to external perturbations [6].
The Impedance Controller has two modes of
operation. The first one is in Cartesian space, which sets
the desired impedance at the tool centre point (TCP).
The second one is at joint level, for setting individual
desired joint impedances. In both cases, stiffness and
damping matrices are predefined, i.e., they are
controller parameters which define the compliant
behaviour [7].
Cartesian impedance control achieves a desired
dynamical relationship, the impedance, between
external forces and movements of the robot. This is
important for the WM in order to cope with possible
inaccuracies and to ensure compliant interaction with
the environment. This is possible thanks to the torque
interface available in the WM. Indeed, the Cartesian
impedance control takes as input a desired position and
orientation in the Cartesian space, often referred to as
virtual equilibrium, and computes the generalized 6
degrees of freedom forces (i.e. three translational forces
and three rotational torques). By applying the Jacobian
transformation, these generalized forces are then
transformed into joint space, a torque vector with seven
elements. This torque vector is the input to the joint
torque controllers. The torque controller ensures that the
desired torque is commanded into each joint. By virtue
of the torque sensors, located at each joint, inertia
shaping is possible. Indeed, thanks to the torque
controller, the apparent inertia of each joint is scaled
down, thus maximizing the performance of the robot.
Besides the torque control mode, the WM can also
be controlled in position mode, which is mainly used for
scenarios where the robot moves freely in space,
without contacts with the environment.
The general control architecture for the robot is
shown on Fig. 12; it includes all the components to
control the walking manipulator (WM) from top-level
(planning and command generation) to low-level
control. Each joint of the WM is equipped with a local
controller for the closed loop position/current control of
the actuator and the measurements of the joint sensors.
All joint controllers are interfaced through an EtherCAT
bus (master/slave), which ensures the real-time
exchange of information required for the control
algorithm of the arm at 500 Hz (minimum).
Page 8 of 10
The WM controller communicates with the WM via
an encapsulated data layer, which hides the specific
communication details (i.e. the Master I/F and the CAN
Driver). That means, that data packets are interfaced to
the appropriated communication bus using an internal
API (IP-based) to local SW processes running close to
the OS/Driver layer and in charge of the encoding of the
high-level information in the appropriated low-level
format. The data communication between the WM
controller and the EtherCAT Master must fulfil the real-
time requirements, given by the control rate of the WM.
As both software modules are running on the WMC, the
implemented interface is based on the UDP/IP protocol.
Fig. 12 also shows the interface to the planner. At
the left, the planning component provides a coarse,
collision-free trajectory to the WMC, which is then
interpolated and commanded to the actuators at high
rate. Therefore, the interface between the planner and
the WMC is characterized by an asynchronous
command mode: the planner sends commands as a
queue of events, according to the current execution state
of the WMC. It has to act as a state machine, which fires
new commands if the on-going command has been
finished. The command interface from the Planner to
the WMC mainly includes two operational commands:
A transfer motion between the current joint
angle configuration to a desired one, described
by a list of joint configurations. This motion is
fully expressed in the joint space of the
manipulator arm and is performed by an
interpolator, which handles the intermediate
via points in a continuous way, controlled by a
simple position controller.
An approach/docking motion, which is
expressed in the Cartesian space (tool centre
point of the manipulator w.r.t. a reference
coordinate system, e.g. the local reference of
the servicer satellite). This motion results in a
straight-line impedance-controlled motion
from the last (reached) pose of the transfer
motion to the desired alignment pose. We
assume that this Cartesian motion will not
drive the manipulator into the limits of its
workspace.
Both actions give continuous feedback to the planner
about the current execution state.
7. Primary Results
As part of the detailed design and early MAIT
phases of the MOSAR project, one elbow joint (Fig. 13)
and the HOTDOCK interfaces (Fig. 15) have been
manufactured, assembled, integrated and tested
successfully. They have been interfaced with the
avionics components to validate the first version of the
control software.
The elbow joint prototype has been integrated in a
dedicated 3D printed prototype casing, see Fig. 14.
Early tests of position and torque controls have also
been successfully performed to validate the accuracy
and compliance behaviours or the motor joint.
In parallel to the development of the MOSAR-WM
motorization, successful integration of HOTDOCK
standard interfaces has been performed (see Fig. 15),
validating the full functionality of the device when
executing the docking and deploying the central
functional interface [8].
Fig. 12. Robot control architecture
Page 9 of 10
Fig. 13. MOSAR-WM Robot Joint.
Fig. 14. MOSAR-WM elbow prototype.
Fig. 15. HOTDOCK Standard Interface.
8. Conclusion
This paper introduced MOSAR-WM, a relocatable
robotic manipulator for future on-orbit applications. In
comparison to other robotic manipulator concepts for
future on-orbit applications, MOSAR-WM tackles
modular missions related to assembly and
reconfiguration of satellites, coupled with the paradigm
of standardisation of spacecraft modules. MOSAR-WM
benefits from an innovative multidisciplinary design
(mechanics, avionics, electronics, software and control)
for performing manipulation tasks as well as relocating
itself over the spacecraft structure. First MAIT of
mechatronic subassemblies and key technologies of the
MOSAR-WM, motor joint and HOTDOCK standard
interface, have been successfully performed. Future
work will be focused on performing the complete
hardware and software integration as well as testing the
MOSAR-WM within the MOSAR demonstrator. In
parallel to this activity, the approach of relocatable
robotic systems using standard interconnects is extended
through the ESA project MIRROR, depicted in Fig. 16,
where a multi-arm installation robot for on-orbit large
assembly is being developed.
Fig. 16. Artist representation of MIRROR primary
concept.
Acknowledgement
This work was partially funded by the European
Commission Horizon 2020 programme under grant
number 821966 (MOSAR).
References
[1] M.A. Roa, K. Nottensteiner, A. Wedler, G.
Grunwald. Robotic Technologies for In-Space
Assembly Operations. In: Symposium on Advanced
Space Technologies in Robotics and Automation –
ASTRA, 2017.
[2] JP. Davis, JP. Mayberry, JP. Penn. On-orbit
Servicing: Inspection, Repair, Refuel, Upgrade, and
Assembly of Satellites in Space, 2019.
[3] A. Rusconi, P. Magnani, P. Campo, et al. Dexarm
engineering model development and test. In 10th ESA
Workshop on Advanced Space Technologies for
Robotics and Automation-ASTRA, 2009.
Page 10 of 10
[4] A. Beyer, G. Grunwald, M. Heumos, et al. CAESAR:
Space robotics technology for assembly, maintenance,
and repair. In 69th International Astronautical Congress
(IAC), Bremen, Germany, 2018.
[5] P. Letier, X. Yang, M. Deremetz et al., MOSAR:
Modular Spacecraft Assembly and Reconfiguration
Demonstrator, ASTRA, 15th Symposium on Advanced
Space Technologies in Robotics and Automation,
Noordwijk, Netherlands, 2019.
[6] F. Lange, G. Grunwald, A. Albu-Schäffer. Robotic
simulation of on-orbit servicing including hard impacts.
In: IEEE/RSJ Int. Conf. Intelligent Robots and
Systems – IROS 2016, pp. 2992-2997.
[7] A. Albu-Schäffer, C. Ott and G. Hirzinger. A unified
passivity-based control framework for position torque
and impedance control of flexible joint robots, Int.
Journal of Robotics Research, vol. 26, pp. 23-39, 2007.
[8] P. Letier, T. Siedel, M. Deremetz, E. Pavlovski, B.
Lietaer, K. Nottensteiner, M.A. Roa, J.S. Garcia-
Casarrubios, J.L.C. Romero, J. Gancet. HOTDOCK:
Design and Validation of a New Generation of Standard
Robotic Interface for On-Orbit Servicing. 71th
International Astronautical Congress (IAC), The
CyberSpace Edition, 2020.