On-orbit Free-floating Manipulation using a Two-arm Robotic

System

Jose Luis Ramon

, Jorge Pomares

and Leonard Felicetti

University of Alicante, Department of Physics, Systems Engineering and Signal Theory, Alicante, Spain

Cranfield University, School of Aerospace, Transport and Manufacturing, Cranfield, U.K.

Keywords: Dual-arm Manipulator, Space Robotics, On-orbit Servicing, Visual Servoing, Impedance Control.

Abstract: A direct visual-servoing algorithm is proposed for the control of a space-based two-arm manipulator. The

scenario under consideration assumes that one of the arms performs the manipulation task while the second

one has an in-hand camera to observe the target zone of manipulation. The algorithm uses both the camera

images and the force/torque measurements as inputs to calculate the control action to move the arms to

perform a manipulation task. The algorithm integrates the multibody dynamics of the robotic system in a

visual servoing framework that uses de-localized cameras. Impedance control is then used to compensate for

eventual contact reactions when the end effector touches and operates the target body. Numerical results

demonstrate the suitability of the proposed algorithm in specific tasks used in on-orbit servicing operations.

1 INTRODUCTION

Space manipulators will be used in a growing range

of missions to assemble, repair, resupply, and rescue

satellites in orbit or remove them at the end of life

(Flores-Abad et al., 2014). Rigorous requirements in

terms of accuracy and safety of the robotic operations

are generally imposed in such kinds of applications,

as it is imperative to avoid collisions that can damage

equipment or compromise the success of missions

(Felicetti et al., 2016). These tasks become even more

challenging when targets are in uncontrolled non-

cooperative conditions (Moghaddam & Chhabra,

2021) or the servicing spacecraft is maintained in a

free-floating condition (Xu et al., 2020).

In such kinds of robotic operations, the knowledge

of the relative position between the service spacecraft

and the target must be continuously monitored to

avoid collisions (Cassinis et al., 2019). Onboard

cameras are preferred over all other sensors, as they

have higher technology readiness levels, a higher

degree of reliability, and better versatility than other

solutions (Palmerini et al., 2016). Cameras can be

located on the main body of the servicing spacecraft

or on movable and reconfigurable appendages to

https://orcid.org/0000-0002-8635-6061

https://orcid.org/0000-0002-7523-9118

https://orcid.org/0000-0001-6830-4308

avoid occlusions of the observed scene during the

manipulation task (Peng et al., 2021; Wang et al.,

2017).

This paper focuses on the specific scenario where

a servicing spacecraft equipped with robotic arms

performs on-orbit servicing and manipulation

operations. Specifically, the spacecraft is assumed to

be equipped with two robotic arms serving the

manipulation and observation functions, respectively.

The first manipulator is an anthropomorphic robotic

arm with all the useful tools to grasp and manipulate

parts of the target satellite. The second manipulator is

a robotic arm with an eye-in-hand camera system

used to observe the specific target area where the

robotic operations are performed. In addition, in order

to better perform the manipulator task, it is necessary

to have also knowledge of the contact forces and

torques with the targets. For this reason, the scenario

assumes that force sensors at the end-effector are able

to measure these actions (Garcia et al., 2019).

This paper proposes a new approach to build a

visual servoing controller that considers the relative

free-floating condition of the bodies involved in the

operations and integrates it with an impedance control

strategy for compensating eventual contact reactions

Ramon, J., Pomares, J. and Felicetti, L.

On-orbit Free-ﬂoating Manipulation using a Two-arm Robotic System.

DOI: 10.5220/0010712100003061

In Proceedings of the 2nd International Conference on Robotics, Computer Vision and Intelligent Systems (ROBOVIS 2021), pages 57-63

ISBN: 978-989-758-537-1

during the manipulation. Visual servoing is a well-

known approach to guide robots using visual

information obtained from cameras (Chaumette &

Hutchinson, 2006). Image-based visual servoing

systems allow for the robot guidance by only using

image information and do not require reconstructing

the 3D position of the target to guide the robot. The

proposed controller assumes that the target trajectory

is defined directly in the image space, and it

calculates the torques to be applied to the

manipulator's joints to perform the manipulation task.

The proposed direct image-based visual servoing

control outputs the joint torques directly, without

having internal control loops of servo motors. This

characteristic offers advantages in the guidance of

space robots: eventual actions on the main servicing

spacecraft body can be easily computed and

predicted. In (Alepuz et al., 2016), a direct image-

based controller is proposed for the guidance of a

free-floating manipulator using an eye-in-hand

camera. In (Pomares et al., 2018), a visual servoing

system is proposed to guide a spacecraft during a

rendezvous manoeuvre. In this case, a camera is

attached to the servicing spacecraft. The approach

presented in this paper assumes that the camera is

moving alongside the second arm. Consequently,

control actions are generated independently of the

unknown position of the camera. The second arm can

be moved to ensure better views of the observed scene

of manipulation.

The proposed visual servoing algorithm also

integrates force and torque measurements at the

manipulator end effector to increase the system

robustness when interacting with the target body. A

mix between a visual servoing controller and

impedance control is proposed in (Garcia et al., 2020)

to perform spacecraft docking in on-orbit servicing

operations. Impedance control is also used in (Mitros

et al., 2017) to evaluate and compensate the interface

contact actions during on-orbit docking manoeuvres

between two spacecraft. A servicing spacecraft with

a force-controlled manipulation system is also

presented in (Dalyaev et al., 2018), using an

impedance control to touch and move the end effector

over the target surface. In on-orbit servicing

applications, the relative dynamics and eventual

contact dynamics between the target and the servicing

spacecraft might strongly affect the performance of

the robotic operations. In such circumstances, indirect

approaches are preferable over direct ones. This is the

case of the algorithms proposed in (Garcia et al.,

2019) (Garcia et al., 2020), where the impedance

controller is paired with a vision control to help keep

the manipulator aligned during the manipulation

tasks. The algorithm presented in this paper proposes

an image-based visual impedance control law that

simultaneously combines the inputs from the camera

and the force sensors. Simulation results will show

that the use of such a controller allows for an increase

of tracking precision with respect to the previous

direct visual servoing approaches. To validate the

methodology, this paper shows the results of an

insertion task. The proposed controllers can

compensate for the tool's eventual misalignments

while this is inserted in a hole in the target spacecraft.

The remaining part of the paper is organized as

follows: Section 2 describes the system architecture

proposed for the servicing spacecraft and its

dynamics. The visual servoing and interaction control

are described in Section 3 and Section 4, respectively.

Section 5 presents the numerical results used to assess

the controllers' validity and robustness on the

trajectory tracking and the tool insertion tasks.

Finally, Section 6 summarizes the main findings and

presents the concluding remarks.

2 SYSTEM ARCHITECTURE

2.1 On-orbit Servicing Scenario

A representation of the on-orbit servicing scenario is

shown in Figure 1. A servicing spacecraft is supposed

to perform some on-orbit servicing operations to a

target spacecraft. The servicing spacecraft is equipped

with two robotics arms that serve two functions:

manipulation and observation, respectively. The first

manipulator has 𝑛𝑒 rotational joints and it is used for

accomplishing the manipulation tasks. The second

manipulator is a robotic arm with 𝑛𝑐 degrees of

freedom with an eye-in-hand camera system.

Figure 1: On-orbit servicing scenario.

X(m)

Y(m)

Z(m)

Base

Spacecraf

Robotic

camera

Manipulato

Targe

Spacecraf

Target

features

ROBOVIS 2021 - 2nd International Conference on Robotics, Computer Vision and Intelligent Systems

A pattern with 𝑚 points is attached to the target

spacecraft. The robotic camera will use this to

identify the relative pose of the target spacecraft.

Figure 1 also shows the coordinate frames

adopted in this study. The B frame is attached to the

servicing spacecraft's main body, and the T frame is

used for the target spacecraft. Two other coordinate

frames represent the pose of the end-effectors of both

the robotic arms: the C frame at the eye-in-hand

camera and the E frame at the end of the manipulator's

arm. Finally, an Earth Centered Inertial coordinate

frame, called I, is adopted as a reference frame for

calculating the objects' positions and attitudes

included in the scenario.

2.2 System Dynamics

The current configuration of the servicing spacecraft

and its two robotic arms is represented by the state

vector 𝝐𝒕







,𝝓







,𝒒





,𝒒









, where 𝒕





and 𝝓





are

the position vector and attitude coordinates (Euler

angles) of the base spacecraft with respect to the

Inertial frame and, 𝒒





and 𝒒





are the joint angles of

both the manipulator and the robotic camera,

respectively.

The equation of motion of robotic system can be

written as (Pisculli et al., 2014):



𝒉





𝝉



𝝉



= 

𝑴



𝑴



𝑴



𝑴





𝑴



𝑴





0𝑴





𝒗







𝒒





𝒒





+ 

𝒄



𝒄



𝒄





𝑱





𝟎



𝒉





𝟎





(1)

where 𝒒





and 𝒒





is the set of joint accelerations of the

robot manipulator and camera, respectively, 𝒗









𝒕









,𝝎













∈ ℜ

denotes the linear and angular

accelerations of the base spacecraft expressed in the

Inertial coordinate frame, 𝑴



∈ ℜ

6×6

is the inertia

matrix of the base spacecraft, 𝑴



∈ ℜ

6×ne

is the

coupling matrix between the spacecraft and the

manipulator, 𝑴



∈ ℜ

ne×ne

is the inertia matrix of the

manipulator, 𝑴



∈ ℜ

6×nc

is the coupling matrix

between the spacecraft and the robotic camera,

𝑴



∈ ℜ

nc×nc

is the inertia matrix of the robotic

camera; 𝒄



, 𝒄



, and 𝒄



∈ ℜ

are a

velocity/displacement-dependent, non-linear terms

for the base, manipulator and robotic camera,

respectively, 𝒉





∈ ℜ

includes both the force and

torque exerted on the base of the servicing spacecraft

but in this paper no forces and torques will be applied

to the servicing spacecraft, 𝝉



∈ ℜ

and 𝝉



∈ ℜ

are the applied set of joint torques acting on the robot

manipulator and on the robotic camera, respectively.

It is also worth noting that eventual external wrenches

𝒉





on the end effector can be projected into the joint

space by using the Jacobian 𝑱



and therefore can be

included into the robot dynamics.

3 VISUAL SERVOING

The proposed visual-based control uses features of

the target body for driving both the robotic

manipulator and the robotic camera. Figure 1 shows

a pattern of 𝑚 points attached on the body of the

target satellite that might represent possible visual

features observed by the robotic camera. These points

have fixed positions with respect to the target

coordinate frame (𝒑

,





𝑥

,



𝑦

,



𝑧

,



, with 𝑖

1…𝑚), but they will appear as 2D points in the

camera image plane 𝒔

,

𝑋

,

,𝑌

,





∈ℜ



, after

being projected through a pin-hole camera model (Ma

et al., 2015). The controller is built upon the concept

that the same set of 𝑚 features seen in the target can

be virtually generated and attached to the

manipulator's end effector and therefore moving

rigidly with it.

The visual-servoing controller aims to match the

virtual features with ones attached to the target. In this

way, the robotic manipulator follows a specified

trajectory defined in the image plane of the robotic

camera. Thus, the position of each of the virtual

features, 𝒑

,



, with 𝑖1…𝑚, will be considered

constant with respect to the coordinate frame of the

end-effector. The corresponding virtual image

features, 𝒔

,

𝑋

,

,𝑌

,





∈ℜ



, are obtained taking

into account the manipulator kinematics. The camera

position is known from the actual arm configuration;

therefore, it is possible to relate the manipulator end-

effector position with the position of the robotic

camera through an algebraic relation given by the

direct kinematics of the two manipulators. A pin-hole

camera model is then used for projecting each of the

points in the camera frame 𝒑

,



𝑥

,



,𝑦

,



,𝑧

,







onto the image plane, 𝒔

,

𝑋

,

,𝑌

,





, using the

following equation:

𝒔

,



𝑧

,





𝑥

,



𝑦

,





(2)

where 𝒑

,



𝑥

,



,𝑦

,



,𝑧

,







∈ℜ



is the position of

the i-th point with respect the camera frame. The time

derivative of the virtual features are:

𝒔



,

𝑋



,

,𝑌



,





𝑳

,

𝒑



,



(3)

On-orbit Free-ﬂoating Manipulation using a Two-arm Robotic System

with:

𝑳

,



𝑧

,





10𝑋

,

01𝑌

,



(4)

The value of 𝒑



,



can be obtained from 𝒑

,



taking

into account the relationship between both frames:

𝒑



,





𝑬



𝑠𝑘𝑹





𝒑

,







𝑹





𝟎



𝟎



𝑹









∙



𝒗





𝒗







(5)

where 𝑹





is the rotation matrix between the camera

and the Inertial frame, 𝒗





and 𝒗





are the twist of the

end-effector, E, and the camera, C, with respect the

Inertial frame, and 𝑬



∈ℜ



the identity matrix.

Finally, the Jacobian matrix 𝑱

,

can be defined as:

𝒔



,

𝑳

,



𝑬



𝑠𝑘𝑹





𝒑

,







𝑹





𝟎



𝟎



𝑹











𝒗





𝒗









𝑱

,



𝒗





𝒗







(6)

On the other hand, the time derivative of the visual

features extracted from the target spacecraft (using

the robotic camera) can be obtained using the

interaction matrix, 𝑳

,

, used in classical image-based

visual servoing systems ( Ma et al., 2015):

𝑳

,



⎣

⎢

⎡



𝑧

,



𝑋

,

𝑧

,



0

𝑧

,



𝑌

,

𝑧

,



𝑋

,

𝑌

,

1  𝑋

,



𝑌

,

1  𝑌

,



𝑋

,

𝑌

,

𝑋

,



(7)

Therefore:

𝒔



,

𝑋



,

,𝑌



,





𝑳

,



𝑹





𝟎



𝟎



𝑹









𝒗







𝑱

,

𝒗

(8)

The aim of the visual-servoing controller is to reduce

the image error 𝒆



𝒔



𝒔



to zero, where 𝒔







𝒔



,𝒔



,…,𝒔







and 𝒔







𝒔



,𝒔



,…,𝒔







are

the virtual features and the real ones extracted by the

robotic camera, respectively.

4 INTERACTION CONTROL

The proposed control scheme takes the contact

dynamics between the two bodies into account to

compensate for eventual reactions and disturbances

produced during the contact. This study assumes that

the target spacecraft has a greater mass than the

servicing spacecraft so that the target's motion does

not change significantly due to the interaction with

the servicer's end-effector. On the other hand,

reaction forces produced by the contact dynamics can

produce significant changes in both the position and

attitude dynamics of the servicing spacecraft.

A damper-spring model is used in this paper to

characterize this kind of interaction. Therefore, the

visual servoing approach generates a virtual damper-

spring behaviour for the pose displacement generated

by the visual error (Tommasino et al., 2020)

𝑫𝒗





𝜶𝒉





(9)

where 𝑫 is the damping matrix, 𝒗





is the desired

twist of the manipulator-end and includes both linear

and angular velocities, 𝒉





∈ℜ



is the external

wrench action on the manipulator and 𝜶∈ℜ



is the

control law to be defined for the visual servoing task.

The following Lyapunov function is considered:

𝑽



𝒆







𝒆





𝑸𝒆



(10)

where 𝑸 is a diagonal positive definite matrix to

guarantee the system stability. The time derivative of

the previous Lyapunov function is:

𝑽





𝒆





𝒆





𝑸𝒆





(11)

By using Eq. (6) and (8), the time derivative of the

image error can be calculated as:

𝒆





𝒔





𝒔







𝑱





𝒗





𝒗









𝑱



𝒗







𝑱



𝒗









𝑱





𝑱





𝒗







𝑱



𝒗







𝑱



𝒗





(12)

where 𝑱



𝑱

,



,𝑱

,



,…,𝑱

,







∈ℜ



, 𝑱





𝑱

,



,𝑱

,



,…,𝑱

,







∈ℜ



and 𝑱



𝑱



𝑱



Thus, by taking into account Eq.(12) and (9), the

value of 𝑽



becomes:

𝑽





𝒆





𝒆





𝑸



𝑱



𝒗







𝑱



𝒗







𝒆





𝑸



𝑱



𝑫





𝒉





𝜶





𝑱



𝒗







(13)

and, if we consider the following control action:

𝜶𝑫



𝑲

𝑱





𝑸𝒆





𝑱





𝑱



𝒗







(14)

being 𝑲 a positive definite matrix, Eq. (13) becomes:

𝑽





𝒆





𝒆





𝑸

𝑱



𝑫



𝒉





𝒆





𝑸

𝑱



𝑲

𝑱





𝑸𝒆



(15)

It is worth noting that, when the robot does not

interact with the target spacecraft, there are no

external actions acting on the end-effector (𝒉





0),

therefore 𝑽





𝒆





𝒆





𝑸𝑱



𝑲𝑱





𝑸𝒆



and

consequently, the system in Eq.(9) is asymptotically

ROBOVIS 2021 - 2nd International Conference on Robotics, Computer Vision and Intelligent Systems

stable: the reference trajectory tracking task will be

achieved at the equilibrium 𝒆



0 if 𝑱



nonsingular. On the other hand, when the robot

interacts with the target spacecraft, Eq. (9) can be

used in conjunction with Eq.(14) to obtain:

𝑫𝒗





𝑫𝑲



𝑱





𝑸𝒆



𝑲



𝑱





𝑱



𝒗







𝒉





(16)

The desired interaction compliance can be defined in

the Cartesian space by means of the matrices 𝑲 and

𝑫 and the image convergence can be regulated by

setting selected gains for 𝑸. Specifically, this is done

by formulating an optimal control strategy for

tracking the reference trajectory obtained from the

interaction wrench and the image error. This tracker

has been developed in (Pomares et al., 2015).

5 RESULTS

Simulations have been carried on to assess the

performance of the proposed visual-servoing control

strategy when performing the task of tool insertion

into the body of the target spacecraft. A predefined

trajectory is followed by the robotic camera. This

trajectory is pre-planned offline so that the visual

features remain within the field of view during the

task. The control of the robot camera trajectory is out

of the scope of this study, but approaches like the ones

shown in (Garcia et al., 2020) (Mitros et al., 2017) can

guarantee that the camera can correctly observe the

areas of interest for performing the operations.

The mass properties of the manipulator, of the

robotic camera and of the base satellite are listed in

Table 1. The controller matrices are set as follows:

𝑸𝑑𝑖𝑎𝑔



0.01



∈ ℜ

2m2m

𝑲 = 𝑑𝑖𝑎𝑔0.1,0.1,0.4,10,10,15,

𝑫𝑑𝑖𝑎𝑔100,100,400,10,10,20,

where 𝑑𝑖𝑎𝑔  is a matrix

with the diagonal elements

equal to the argument of the function. The camera

acquires 20 images per second with a resolution of

640x480 pixels but the control loop is running at 5

ms. The camera is supposed to be previously

calibrated and the intrinsic parameters are (u

, v

) =

(298, 225) px, and (f

, f

) = (1082.3, 1073.7) px,

where u

and v

are the position of the optical center

and f

are the focal lengths in the x and y

directions, respectively. A tool is held by the robotic

manipulator end-effector and needs to inserted 2 cm

into the target spacecraft.

An offset of 3 mm from the ideal configuration of

the virtual features is included in the desired pattern

to simulate an error in the final pose. In this way, it is

possible to evaluate the effectiveness of the proposed

control scheme to guide the robotic manipulator in the

presence of contact forces: some adjustments will be

needed by the embedded impedance control.

Table 1: Dynamic parameters of the robot.

Base

Mass

(Kg)

Inertia (kg∙m

)

2550 6200 3540 7090

Arms

Mass

(Kg)

Inertia (kg∙m

)

2550 6200 3540 7090

Link1

35 2 0.2 2

Link2

22 3 0.2 3

Link3

22 3 0.2 3

Link4

10 0.15 0.2 0.4

Link5

10 0.15 0.2 0.3

Link6

10 0.2 0.25 0.3

The simulation results are shown from Figure 2 to

Figure 4. Specifically, the 3D trajectory described by

the robotic system is shown in Figure 2, where the

trajectory of the manipulator is highligthed in red and

the trajectory of the robotic camera in blue. From the

overlapping frames, it is possible to evaluate the

movements of the manipulator, which tries to extend

its arm to reach the target. At the same time, the

floating base of the satellite moves backwards as a

reaction to the motion of both manipulator and

robotic camera.

Figure 2: Robot arm trajcetories during the tool insertion

task.

The corresponding motion of the image features

in the image plane is shown in Figure 3. The

trajectories of the visual features extracted from the

target spacecraft are represented in blue, and the

trajectories of the virtual features are shown in red.

On-orbit Free-ﬂoating Manipulation using a Two-arm Robotic System

Empty circles and the final features indicate the initial

features are shown by solid circles.

Figure 3: Image trajectories of the virtual and real visual

features.

Figure 4: Reaction forces and torques during the insertion

task.

As expected, the positions of the extracted image

features and virtual features are initially very far from

each other, but they tend to approach each other

during the manoeuvre. However, their final position

in the image plane is not perfectly matching due to the

offset between the actual virtual target configuration

and the ideal one. The controller, in any case,

compensates for this error by using the impedance

strategy shown in Section 4. The time behaviour of

the reaction forces and torques at the end effector of

the manipulator are shown in Figure 4. The peak of

the contact forces is reached after 7 s, when the tool

touches the target for the first time but cannot be

correctly inserted in the first attempt. After this initial

phase, the offset on the virtual target image is

compensated within the controller. The contact forces

reduce their values when the tool is centred and

inserted into the hole.

6 CONCLUSIONS

The paper presented a direct visual servoing

algorithm suitable for on-orbit servicing and

manipulation. The algorithm is applicable to a

spacecraft equipped with two-arm manipulator. The

two arms are dedicated to manipulation and

observation tasks, respectively.

A visual servoing controller independent from the

observed scene's point of view was consequently

developed. The virtual features could be virtually

reconstructed following a specific pattern seen on the

target body and consequently assumed attached to the

end effector of the operating manipulator.

The controller was able to drive the manipulator

in such a way to make the virtual features match the

real features on the target body. Under an impedance

control scheme, the controller also compensated for

eventual contact reactions between the end effector

and the target satellite. Simulations demonstrated the

applicability of this scheme in a standard multi-

degrees of freedom manipulator scenario where

eventual misalignments that would not allow for a

tool insertion task inside the body of the target

satellite were compensated and corrected by the

controller, making this kind of operation still

successful.

Further studies will assess the robustness of the

proposed controller against environmental torques

and forces as well as will evaluate the performance of

the controller with different frame rates of the camera,

and will compare the results with other tracking

controllers.

REFERENCES

Alepuz, J. P., Emami, M. R., & Pomares, J. (2016). Direct

image-based visual servoing of free-floating space

manipulators. Aerospace Science and Technology, 55,

1–9. https://doi.org/10.1016/j.ast.2016.05.012.

Cassinis, L. P., Fonod, R., & Gill, E. (2019). Review of the

robustness and applicability of monocular pose

estimation systems for relative navigation with an

uncooperative spacecraft. Progress in Aerospace

Sciences, 110, 008.

Chaumette, F., & Hutchinson, S. (2006). Visual servo

control. I. Basic approaches. IEEE Robotics &

ROBOVIS 2021 - 2nd International Conference on Robotics, Computer Vision and Intelligent Systems

Automation Magazine, 13(4), 82–90. https://doi.org/

10.1109/MRA.2006.250573.

Dalyaev, I., Titov, V., & Shardyko, I. (2018). A concept of

robotic system with force-controlled manipulators for

on-orbit servicing spacecraft. En Proceedings of the

Scientific-Practical Conference «Research and

Development - 2016» (pp. 239-245). Springer

International Publishing.

Felicetti, L., Gasbarri, P., Pisculli, A., Sabatini, M., &

Palmerini, G. B. (2016). Design of robotic manipulators

for orbit removal of spent launchers' stages. Acta

Astronautica, 119, 118–130. https://doi.org/10.1016/

j.actaastro.2015.11.012.

Flores-Abad, A., Ma, O., Pham, K., & Ulrich, S. (2014). A

review of space robotics technologies for on-orbit

servicing. Progress in Aerospace Sciences, 68, 1–26.

https://doi.org/10.1016/j.paerosci.2014.03.002.

Garcia, J., Gonzalez, D., Rodriguez, A., Santamaria, B.,

Estremera, J., & Armendia, M. (2019). Application of

Impedance Control in Robotic Manipulators for

Spacecraft On-orbit Servicing. 2019 24th IEEE

International Conference on Emerging Technologies

and Factory Automation (ETFA), 836–842.

https://doi.org/10.1109/ETFA.2019.8869069.

Garcia, J., Rodriguez, A., Estremera, J., Santamaria, B.,

Gonzalez, D., & Armendia, M. (2020). Visual Servoing

and Impedance Control in Robotic Manipulators for

On-Orbit Servicing. 2020 25th IEEE International

Conference on Emerging Technologies and Factory

Automation (ETFA), 1, 734–741. https://doi.org/

10.1109/ETFA46521.2020.9211989.

Ma, G., Jiang, Z., Li, H., Gao, J., Yu, Z., Chen, X., Liu, Y.-

H., & Huang, Q. (2015). Hand-eye servo and

impedance control for manipulator arm to capture target

satellite safely. Robotica, 33(4), 848–864.

https://doi.org/10.1017/S0263574714000587.

Mitros, Z., Rekleitis, G., & Papadopoulos, E. (2017).

Impedance control design for on-orbit docking using an

analytical and experimental approach".

Moghaddam, B. M., & Chhabra, R. (2021). On the

guidance, navigation and control of in-orbit space

robotic missions: A survey and prospective vision. Acta

Astronautica, 184, 70–100. https://doi.org/10.1016/

j.actaastro.2021.03.029.

Nocerino, A., Opromolla, R., Fasano, G., & Grassi, M.

(2021). LIDAR-based multi-step approach for relative

state and inertia parameters determination of an

uncooperative target. Acta Astronautica, 181, 662–678.

https://doi.org/10.1016/j.actaastro.2021.02.019.

Palmerini, G. B. (2016). Relative navigation in autonomous

spacecraft formations. 2016 IEEE Aerospace

Conference, 1–10. https://doi.org/10.1109/AERO.20

16.7500944.

Peng, J., Xu, W., Liu, T., Yuan, H., & Liang, B. (2021).

End-effector pose and arm-shape synchronous planning

methods of a hyper-redundant manipulator for

spacecraft repairing. Mechanism and Machine Theory,

155, 104062–. https://doi.org/10.1016/j.mechmach

theory.2020.104062.

Pisculli, A., Felicetti, L., Gasbarri, P., Palmerini, G. ., &

Sabatini, M. (2014). A reaction-null/Jacobian transpose

control strategy with gravity gradient compensation for

on-orbit space manipulators. Aerospace Science and

Technology, 38, 30–40. https://doi.org/10.1016/j.ast.20

14.07.012.

Pomares, J., Felicetti, L., Pérez, J., & Emami, M. R. (2018).

Concurrent image-based visual servoing with adaptive

zooming for non-cooperative rendezvous maneuvers.

Advances in Space Research, 61(3), 862–878.

https://doi.org/10.1016/j.asr.2017.10.054.

Pomares, J., Jara, C. A., Pérez, J., & Torres, F. (2015).

Direct visual servoing framework based on optimal

control for redundant joint structures. International

Journal of Precision Engineering and Manufacturing,

16(2), 267–274. https://doi.org/10.1007/s12541-015-

0035-z.

Tommasino D., Cipriani G., Doria A., Rosati G. (2020)

Effect of End-Effector Compliance on Collisions in

Robotic Teleoperation. Applied Sciences. 10(24), 9077.

https://doi.org/10.3390/app10249077.

Wang, H., Guo, D., Xu, H., Chen, W., Liu, T., & Leang, K.

K. (2017). Eye-in-hand tracking control of a free-

floating space manipulator. IEEE transactions on

aerospace and electronic systems, 53(4), 1855-1865.

Xu, R., Luo, J., & Wang, M. (2020). Kinematic and

dynamic manipulability analysis for free-floating space

robots with closed chain constraints. Robotics and

Autonomous Systems, 130, 103548–. https://doi.org/

10.1016/j.robot.2020.103548.

On-orbit Free-ﬂoating Manipulation using a Two-arm Robotic System