An Unmanned Aerial Carrier and Anchoring Mechanism for

Transporting Companion UAVs

Yiyong Gou

, Lucas Dahl

, Jan Kr

uger

, Cavid Karca

, Dean Boonen

and Rico M

ockel

DKE SwarmLab, Department of Data Science and Knowledge Engineering,

Maastricht University, Maastricht, The Netherlands

Keywords:

Unmanned Aerial Carrier, Anchor Design, Companion UAV, Aerial Anchoring Mechanism.

Abstract:

This paper demonstrates an unmanned aerial carrier as well as a new anchoring mechanism for connecting

and transporting companion unmanned aerial vehicles (UAVs). Establishing this platform presents unique

challenges including the requirements of precise localization of the platform, real-time environment mapping

system, robust ﬂight control approach, docking safety mechanism, and reliable anchor system for the compan-

ion UAV. To obtain the positioning information, a tightly-coupled visual-inertial optimization based odometry

is implemented with a ﬁsheye camera and an inertial measurement unit. A 3D map is updated in real-time

using an Octomap framework. A nonlinear position model predictive controller cascaded with a DJI attitude

controller is implemented for the ﬂight control. Innovatively, we designed a lightweight anchoring mechanism

for safe landing and reliable transportation of the companion UAV. Real-world experiments results suggest

that the transportation system is a viable approach to transport the companion UAV, and that the proposed

anchoring mechanism allows for reliable operation.

1 INTRODUCTION

Recent years have witnessed the development and ex-

ploration of a variety of multi-robot systems. In these

distributed systems, robots cooperate to achieve tasks

that an individual robot would not be capable of. Of-

ten the strengths of robots with different conﬁgura-

tions and abilities are combined. Demonstrated exam-

ples of multi-robot systems include multi-drone par-

cel delivery (Peng et al., 2019), self-reconﬁgurable

perception-driven modular robots (Daudelin et al.,

2018), and the combination of unmanned aerial vehi-

cles (UAVs) and unmanned ground vehicles (Yu et al.,

2018; Ullah et al., 2021).

This work contributes the design and exploration

of a new light-weight anchoring mechanism for aerial

carrier UAVs. In addition, an aerial carrier is pre-

sented that, with the help of the proposed anchoring

mechanism, can carry a small companion UAV. The

https://orcid.org/0000-0001-7544-4675

https://orcid.org/0000-0002-6536-2243

https://orcid.org/0000-0002-7589-7604

https://orcid.org/0000-0002-8270-7418

https://orcid.org/0000-0001-8096-1829

https://orcid.org/0000-0001-5497-3754

combination of both UAVs is advantageous for human

operators that e.g., desire to perform aerial inspec-

tions: The aerial carrier provides to the companion

UAV long-distance travel, communication, and con-

trol as well as the necessary computational resources

for automatic image processing. In return, the com-

panion UAV allows reaching small areas that the car-

rier cannot get into. The more expensive companion

UAV can also be chosen to undergo inspection oper-

ations that might lead to a loss of the UAV and thus

are too risky to be handled by the higher-costs carrier

UAV.

1.1 Related Work

In general, aerial robotic transportation platforms re-

quire a reliable onboard ﬂight system which typically

consists of precise localization, environment percep-

tion and ﬂight control modules (Lutz et al., 2020).

Aerial robotic localization has been investigated us-

ing computer vision (Qin et al., 2018; Bloesch et al.,

2017), global navigation satellite system (Loianno

et al., 2018), inertial measurements (Afrisal et al.,

2019), laser (Zhang and Singh, 2014), and Ultra

Wide Band (Perez-Grau et al., 2017) based meth-

ods, among which the multi-sensors fusion approach

Gou, Y., Dahl, L., Krüger, J., Karca, C., Boonen, D. and Möckel, R.

An Unmanned Aerial Carrier and Anchoring Mechanism for Transporting Companion UAVs.

DOI: 10.5220/0010655500003061

In Proceedings of the 2nd International Conference on Robotics, Computer Vision and Intelligent Systems (ROBOVIS 2021), pages 103-112

ISBN: 978-989-758-537-1

103

is often utilized to obtain higher localization pre-

cision. Mostly lightweight lidar and stereo cam-

eras are selected to provide 3D information for en-

vironment perception (Hornung et al., 2013; Lin

et al., 2019). Cascade control techniques with

two level controllers are generally implemented in

ﬂight control application and onboard position con-

troller such as proportional-integral-derivative con-

trol, gain-scheduled proportional-integral-derivative

control(Wasim et al., 2019), model predictive con-

trol (MPC) (Kamel et al., 2017; Kamel et al., 2016;

Tzoumanikas et al., 2019), and geometric control

(Lee et al., 2010) are adopted and normally followed

by an off-the-shelf autopilot as the inner loop attitude

controller.

With respect to the UAV docking platform, an

aerial recharging docking platform was developed

and implemented in (Jain and Mueller, 2020; Jain

et al., 2020) and they are both tested to achieve ac-

curate docking with the localization information pro-

vided by an indoor position tracking system. As-

sisted with artiﬁcial vision, a ground landing mech-

anism with four bowl-shape inner cones surfaces was

developed for the UAV docking and recharging (Coc-

chioni et al., 2014), but it was designed for ground

static landing and cannot ensure safety in case of

the carrier wobbling that might cause the compan-

ion UAV dropping. Moreover, a UAV docking plat-

form with the ground mobile manipulator attached

with an eye-in-hand visual sensor was built and im-

plemented with UAV (Narv

aez et al., 2017; Narv

aez

et al., 2020). A special in-ﬂight docking platform with

a customized vertical mast as a docking mechanism

mounted above on the aerial carrier was simulated

with a quadcopter assisted by the Global Positioning

System (GPS) and a computer vision process (Rocha

and Robinson, 2020). In current research, the dock-

ing platforms are mostly operating on the ground or

lack of the ability of environment perception, or rely

on an external positioning system. Additionally, these

platform are not equipped with robust fault-tolerance

mechanisms which ensure reliable and safe docking.

1.2 Contributions

The contributions of this paper include the following:

1. A novel design and validation of a lightweight an-

chor system which is customized for docking and

transporting companion UAVs.

2. Development and design of hardware and soft-

ware of a custom-made aerial carrier with

mounted anchoring mechanism.

2 PLATFORM HARDWARE

DESIGN

Given the requirement of achieving aerial transport

and anchoring for the companion UAV, the consid-

erations concerning the platform hardware design in-

clude the structural stability, available payload, struc-

tural protection, and fault-tolerance of the anchoring

mechanism and aerial carrier. In this section, we

present the system hardware as shown in Figure 1.

Companion UAV

Anchoring mechanism

Aerial carrier

Figure 1: Aerial carrier with anchoring mechanism and at-

tached companion UAV.

2.1 Hardware Overview

The whole system hardware consists of four major

elements: (1) an aerial carrier with onboard com-

putational resources running software for perception

and control, (2) an anchor mechanism integrated onto

the aerial carrier for anchoring a companion UAV,

(3) a modiﬁed DJI Tello companion UAV with a

changed downward camera, and (4) a ground station

for monitoring and communication with the aerial

platforms. Figure 2 shows a block diagram of the

electronics hardware components of the carrier UAV

and ground station. Table 1 provides additional speci-

ﬁcations. With a total system weight (including aerial

carrier 3.918kg, companion UAV 0.086kg, and an-

choring mechanism 0.146kg) of 4.15kg, the carrier

UAV reaches a ﬂight time of up to eight minutes in

operation mode.

2.2 Design of Anchoring Mechanism

and Protection

The anchoring mechanism shown in Figure 3 is com-

posed of a 3mm-thick foam board featuring a frac-

tal marker (Romero-Ramirez et al., 2018; Romero-

Ramire et al., 2019; Garrido-Jurado et al., 2016) for

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

104

D435i

Fisheye

Camera

IMU

DJI N3 Autopilot

Intel NUC i7

Brushless Motor x 4

U2D2 & Power Hub

DYNAMIXEL-XL330 x 2

& Anchor

Battery 4S 2200mAh

Battery 4S 6200mAh

UBEC 5V

UBEC 12V

PWM

USB 3 USB 2

WIFI

Router

ThinkPad

T470S

UART

5V DC

TTL

ESC x 4

PWM

USB 2

DC 4S

12V DC

DC 4S

Three Phase AC

WIFI

Aerial Carrier

Ground Station

Radiolink

AT9S RC

Radio

Figure 2: Ground station and electronic hardware compo-

nents of aerial carrier.

Table 1: System components and speciﬁcations.

Components Speciﬁcations

DC motor * 4 Turnigy 3542-800kv

Propeller * 4 4.7inch×11inch

ESC * 4 Skywalker 50A

Autopilot DJI N3 (IMU)

Sensing • AR0134 ﬁsheye camera

• Intel RealSense D435i

Servo • Dynamixel XL-330-M228

• Dynamixel U2D2 interface

RC controller RadioLink AT9S

Onboard PC Intel NUC7i7DNBE 16G

Battery • LiPo 4S 2200mAh

• LiPo 4S 6200mAh

Body frame Carbon ﬁber 40cm×40cm

Protection • Carbon ﬁber cage 70cm×70cm

• UV polyethylene nets

Anchor • 3mm foam board 25cm×25cm

• PLA 3D printed arm * 2

automatic detection by and landing of a companion

UAV. After landing on the foam board (not discussed

in this paper) and during aerial transport, the compan-

ion UAV is grabbed and ﬁxated by two custom-made

arms (shown in red in Figure 3) that are being con-

trolled by two lightweight Dynamixel motors. The

custom-made arms are designed to be lightweight,

fault-tolerant and stiff enough to secure the compan-

ion UAV. The total weight of the anchoring mecha-

nism (including foam board, electronics, motors, and

arms) is around 172g. Each arm weighs 15g, and they

are being held on by three nylon screws, respectively.

The inner surface of the arm is sloped to prevent any

mating surface from binding and blocking the mech-

anism.

In addition to the anchoring mechanism, a sup-

portive protection structure composed of carbon ﬁber

rods and UV stabilized polyethylene nets has been de-

signed and attached to the aerial carrier to make the

landing and anchoring process more fault-tolerant and

to prevent any falling object from destroying the aerial

carrier system while providing additional safety mar-

gins for the landing of the companion UAV. The total

weight of this protection structure (Figure 1) is around

413g.

Figure 3: CAD view of the anchoring mechanism installed

above the aerial carrier.

2.3 Sensors and Computation

Resources

To achieve the requirements of the aerial robotic envi-

ronment perception and localization, we choose a 1/3

inch 1.2MP global shutter monochrome camera mod-

ule with a 1.58mm focal length ﬁsheye 185 degrees

FOV lens and mount it facing forwards on the aerial

carrier, moreover we selected the lightweight and

small-size DJI N3 ﬂight controller embedded with

an inertial measurement unit (IMU) module which

can provide 400Hz inertial measurements and gyro-

scope data working with 15Hz synchronized ﬁsheye

images to run the localization algorithm described in

Section 3. We selected the lightweight commercial

stereo camera module Intel RealSense D435i which

is mounted on the upper front of the protection frame

and can output raw point cloud data at 30Hz, RGB

images, IMU data, and depth images for environ-

mental mapping. To allow all aerial carrier soft-

ware running onboard, a powerful mini sized PC In-

tel NUC7i7DNBE is utilized to process all sensors

and actuation data and to communicate with the on-

board autopilot and ground station, depicted in Fig-

ure 2. To monitor the system status and command the

An Unmanned Aerial Carrier and Anchoring Mechanism for Transporting Companion UAVs

105

aerial carrier and companion UAV, a ground station

was equipped with a ThinkPad PC communicating

with the aerial systems via WiFi and a RC controller

to command the system operation via radio signal.

2.4 Power Management

As shown in Figure 2, two 4S LiPo batteries provide

stable power supply for the onboard aerial carrier us-

ing two UBEC modules and the power budget of all

the components are listed in Table 2 which meets the

requirement of the maximum operation time of eight

minutes.

Table 2: System power budget.

Component Input voltage(V) MP(W)

DC motor ESC 14.80 493.00×4

Onboard PC 12.00 59.31

Dynamixel servo 5.00 7.35 × 2

Autopilot 14.80 4.80

Fisheye camera 3.30 0.40

D435i stereo 5.00 2.50

MP: Maximum power.

3 SOFTWARE AND ALGORITHM

DESIGN

In this section, we address the software components

for the aerial carrier platform to achieve transportation

and automatic anchoring operation.

3.1 Software Overview

Figure 4 provides an overview of system software

modules and architecture implemented on the aerial

carrier for stable position ﬂight control, map forma-

tion, and the control of the anchoring mechanism.

The positioning module’s functionality is to pro-

vide the pose information for the mapping algorithm

and feedback for the position controller. Due to this

positioning module and by using its D435i stereo

camera, the aerial carrier is capable of generating 3D

maps of its environment that can be used by a human

operator for further analysis. For stable ﬂight con-

trol e.g., during transportation, a position controller

for the aerial carrier has been developed considering

the platform dynamics. For reliable anchoring of the

companion UAV, a software has been designed and

implemented that allows automatic anchoring. As

shown in Figure 4, all software modules are man-

aged by a ﬁnite state machine (FSM) to accomplish

the transport and anchoring tasks effectively.

D435i

Fisheye Camera

DJI N3 IMU

Trigger

Raw Images

DJI ROS SDK

Raw IMU

Image Timestamper

Trigger-time

Optimization-based VIO with Loop Closure

IMU

Images

D435i ROS

SDK

Images

Octomap

Sever

Point Cloud

Odometry

MPC Position

Controller

DJI N3

Autopilot

Thrust & Attitude

Dynamixel

movement

Intel NUC i7

Dynamixel

controller

Joint Cmd

FSM

controller

S_d

S_r

Visualization

ThinkPad T470s

Figure 4: System software architecture.

3.2 Positioning and Mapping

Precise positioning, being pose feedback information

in position control loop, is a key factor to keep the

aerial carrier stable during transportation and anchor-

ing. Due to the consideration of the platform conﬁg-

uration and our experiment environment, IMU data

and vision data are tightly fused into a visual iner-

tial odometry pipeline to output a reliable and accu-

rate pose (Qin et al., 2018; Qin and Shen, 2018). To

implement this pipeline on this aerial platform, hard-

ware synchronization of the IMU and monocular vi-

sion data is essential for this application and required

by its high accuracy of the positioning. As shown in

Figure 4, hardware synchronization is implemented

by the DJI N3 autopilot IMU that is triggering the

camera image capture at a ﬁxed frequency of 15Hz.

The image data is stamped with the corresponding

IMU timestamp for further processing. Strictly, the

sampling data is not in complete temporal alignment

because of the image exposure and data transmission

delay (Qin and Shen, 2018). However, the time off-

set between IMU and images can be estimated online

along with this optimization based positioning system

(Qin et al., 2018). This positioning system with its

loop closure outputs data at 15Hz including the po-

sition, orientation, linear and angular velocity of the

aerial platform body with respect to the inertia frame

whose origin is aligned with the positioning system’s

initialization point. To increase the frequency of the

localization system, the positioning system output is

directly propagated with IMU data as in (Qin et al.,

2018; Lin et al., 2018) and the ﬁnal system provides

the localization information at 80Hz which is sufﬁ-

cient to be feed back to the position controller. Us-

ing the precise localization information and the point

cloud data from the stereo camera, we employ a map-

ping algorithm Octomap server (Hornung et al., 2013)

to obtain a volumetric 3D occupancy map which is

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

106

applicable to environment perception and path gener-

ation.

3.3 Motion Control

In the aerial robot control ﬁeld, cascaded connection

of a high level controller and a low level controller

is a typical approach to control the aerial robot ﬂy-

ing at a desired pose. We employ an off-the-shelf

autopilot DJI N3 as the low level attitude controller,

which enables the dynamics of the aerial platform to

be approximated. In cooperation with the tuned low

level controller, a position nonlinear MPC controller

with the pose command from RC controller, closely

related to some previous work (Kamel et al., 2017;

Kamel et al., 2016; Tzoumanikas et al., 2019; Car-

los et al., 2020), is developed and implemented on-

board to generate the pitch, roll, and thrust command

for the autopilot while considering external environ-

mental disturbances. Moreover, in the algorithm the

the dynamics of the aerial platform are approximated

with RC data and IMU data, and the pitch and roll

constraints are both set to 20 degrees and thrust com-

mand constraints is set to 120N. An EKF-based dis-

turbance estimator is used within the nonlinear MPC

controller. Subject to the yaw control, a typical P con-

troller is utilized to generate a yaw rate command.

3.4 Anchor Operation

Two servo motors Dynamixel XL330-M288 operate

in position mode and are controlled by the ROS Dy-

namixel workbench which can provide high precision

joint position tracking. In order to operate the arms of

the anchoring mechanism automatically, the ROS ex-

ecution service is called to either drive them smoothly

to their opening or closing position with the motor

speed 1.57rad/s depending on the desired state of the

system with the RC controller triggering.

3.5 Finite State Machine

The behavior of the aerial platform is determined by a

ﬁnite state machine (FSM) consisting of three states:

• RC control: This is the initial state for the aerial

platform taking off, normal landing or emergency

landing. After the take-off is completed and it

is hovering stably, the FSM transitions to the

Transportation state.

• Transportation: The aerial platform is moving

with the command from RC controller. When it

reaches the hover position and hovers stably, the

state switches to Anchor operation . In case of

emergency, the state switches to RC control to

conduct emergence landing.

• Anchor operation: In this state, the anchor is

starting to work for locking or releasing the

companion UAV with a RC switch triggering.

When the action is done, the FSM transitions

to the Transportation state or it transitions to

RC control if an emergency is triggered.

4 EXPERIMENTAL RESULTS

In this section, we conduct multiple experiments to

evaluate the performance of the aerial carrier platform

and the anchor operation for the companion UAV.

4.1 Aerial Transportation

Before implementing the nonlinear MPC controller,

the ﬁrst-order and second-order inner loop attitude

control dynamical model of the aerial platform is

identiﬁed as described in (Sa et al., 2018) to obtain the

approximation of the attitude control loop dynamic.

The model identiﬁcation results are presented in Ta-

ble. 3, and

Table 3: Identiﬁcation of attitude control loop dynamical

model.

Model φ θ

First-order model k

= 1.738 k

= 1.303

= 0.182 τ

= 0.154

Second-order model k

= 1.572 k

= 1.168

= 6.061 ω

= 5.494

= 0.356 ζ

= 0.327

{.}

are the model gains, τ

{.}

are the time con-

stants, ω

{.}

are the damping constants, and ζ

{.}

are

the natural frequencies.

In practice, to conduct aerial transportation, the

ﬂight control of the aerial platform is a critical fac-

tor. To evaluate the stability and effectiveness of the

MPC controller with the onboard positioning module

outputs at 80Hz, the drone is hovering at one way-

point and the hovering result is depicted in Figure 5,

and average error of the x and z axis are both less than

0.05m and the average error of the y axis is less than

0.09m. The MPC controller output results are shown

in Figure 6 and the commands for the attitude con-

troller input are within the range of the attitude and

thrust constraints, and the thrust command is scaled

based on the DJI autopilot input requirement. Addi-

tionally, the aerial carrier is equipped with the ability

to conduct environment mapping which can be used

An Unmanned Aerial Carrier and Anchoring Mechanism for Transporting Companion UAVs

107

for transportation ﬂight path planning. As shown in

Figure 7, the real-time updated 3D occupancy 10cm-

resolution mapping is generated with the OctoMap

framework in a lab environment depicted in Figure

0 20 40 60 80 100

0.35

0.4

0.45

x [m]

odometry ref

0 20 40 60 80 100

-0.1

-0.05

0.05

y [m]

odometry ref

0 20 40 60 80 100

time [s]

0.8

0.85

0.9

z [m]

odometry ref

Figure 5: Hover ﬂight test with odometry.

0 20 40 60 80 100

-0.1

0.1

roll [rad]

0 20 40 60 80 100

-0.05

0.05

0.1

pitch [rad]

0 20 40 60 80 100

time [s]

thrust [N]

Figure 6: Roll, pitch, and thrust command for the aerial

carrier.

Figure 7: 3D occupancy map with OctoMap framework.

4.2 Anchor Test

To test the the effectiveness and reliability of the an-

chor system, ﬁrstly we conﬁgure different numbers

of companion UAV feet grabbed as shown in Figure

9. Subject to each conﬁguration of the number of the

feet grabbed, we do 30 sampling tests with the car-

Figure 8: Lab environment used for ﬂight experiments.

rier wobbling with pitch and roll angles of 90 degrees

respectively, and the results of locking success rate

with different numbers of feet grabbed are presented

in Figure 10 which indicates that with more than 2

feet grabbed the locking performance is reliable and

sufﬁcient for our application.

Figure 9: Grabbed feet conﬁgurations. The feet that are not

held by the arms are highlighted in yellow. a) shows all feet

are successfully grabbed by the arms. c) shows only one

foot being grabbed. e) shows three feet being grabbed. b),

d), and f) present varying two feet conﬁgurations.

With reference to the previous UAV landing re-

search in Table 4, the landing error can be less than

8cm and the corresponding requirement of the dock-

ing platform proﬁle and size of the companion UAV

is included in our design of the docking platform and

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

108

0 1 2 3 4

feet grabbed

20%

40%

60%

80%

100%

locking success rate

100% 100%

96.7%

0 0

Figure 10: Locking success rate with different number of

the feet grabbed.

the choice of the companion UAV.

Table 4: Landing errors of various state-of-the-art UAV

landing methods.

Methods Landing error (cm)

(Yang et al., 2014) 1.5

(Cocchioni et al., 2014) 2.0

(Benini et al., 2016) 2.0

(Yu et al., 2017) 0.4

(Alvarez Custodio, 2019) 7.8

(Sani and Karimian, 2017) 6.0

To obtain the proper docking orientation in yaw

and ﬁnd the better docking success rate, the docking

anchoring experiments on the ground are conducted

when the companion UAV with four anchor feet docks

on the docking fractal marker board with a yaw an-

gle of 0 degrees, 30 degrees, and 45 degrees, respec-

tively, and the coordinate system of the companion

UAV and docking platform are described in Figure 11.

As shown in Figure 12 for experiments with a docking

yaw angle of 0 degrees, the maximum docking center

offset can reach ±8cm with the rotational symmetry

test data mirroring, but only 52.7% of the samples are

grabbed by three or four feet, and based on the results

in Figure 10 and Table 4, it suggests that the compan-

ion UAV can reliably dock with the yaw angle of 0

degrees and offset ±8cm. The experiment results of

docking with a yaw angle of 30 degrees are shown in

Figure 13, and the samples grabbed with three or four

feet occupy around 96.67% but the maximum docking

center offset is only around ±6.5cm. Moreover, the

results of docking with a yaw angle of 45 degrees are

presented in Figure 14 and all the sampling cases are

grabbed with three or four feet, but the the maximum

docking center offset is only around ±6.7cm. Based

on the results of the three cases above, it indicates that

docking with a yaw angle of 45 degrees is relatively

reliable and robust, but its docking available area is

small, which requires higher landing accuracy for the

companion UAV. Specially noted, if the companion

UAV is grabbed with more than two feet during ﬂight

tests, it will be automatically centered by the drone’s

high-frequency shaking and the force applied by the

two servo motors.

yaw

Figure 11: Coordinate system.

-9 -6 -3 0 3 6 9

x [cm]

-9

-6

-3

y [cm]

2 feet

3 feet

4 feet

Figure 12: Companion UAV grab test with yaw angle of 0

degrees.

In mechanic design, we adapt a net protection

structure to ensure the safety of the carrier when the

docking fails on the tag board. We intentionally place

the drone with yaw angle of 0 degrees on sampling

points of the nets protection structure while the drone

is wobbling with pitch and roll angles of 45 degrees

respectively to test its ability to hold the compan-

ion UAV in place in order to demonstrate its fault-

tolerance. The result is displayed in Figure 15.

It suggests that the companion UAV docking with

error of around ±0.35m can be safe and will not de-

stroy the carrier, docking platform or itself.

An Unmanned Aerial Carrier and Anchoring Mechanism for Transporting Companion UAVs

109

-8 -6 -4 -2 0 2 4 6 8

x [cm]

-8

-6

-4

-2

y [cm]

2 feet

3 feet

4 feet

Figure 13: Companion UAV grab test with yaw angle of 30

degrees.

-9 -6 -3 0 3 6 9

x [cm]

-9

-6

-3

y [cm]

3 feet

4 feet

Figure 14: Companion UAV grab test with yaw angle of 45

degrees.

5 CONCLUSIONS

In this paper, we present the system design of an

aerial carrier and anchoring platform for transporting

a companion UAV. To fulﬁll the aerial carrier ﬂight

implementation, an aerial system including precise

localization, environment perception, and ﬂight con-

trol framework was developed and implemented in

an indoor lab environment. Subject to the anchoring

platform, we designed a customized lightweight an-

chor system to lock the companion UAV and a fault-

tolerant protection structure to ensure the docking se-

curity. The effectiveness and reliability of them are

validated by the experiments with different conﬁgu-

rations of feet grabbed and docking orientations. The

future work will focus on the outdoor real-world in-

ﬂight companion UAV docking experiments with our

aerial carrier and anchoring platform.

-40 -20 0 20 40

x [cm]

-40

-30

-20

-10

y [cm]

failure

success

Figure 15: Nets docking fault-tolerance test.

ACKNOWLEDGEMENTS

This research received ﬁnancial support by a Luik-3

grant on advanced brain-robot interfaces.

REFERENCES

Afrisal, H., Rahmadani, T., Nugroho, W. D., Putra, D. A.,

et al. (2019). Inertial navigation system of quadrotor

based on imu and gps sensors. In 2019 6th Interna-

tional Conference on Information Technology, Com-

puter and Electrical Engineering (ICITACEE), pages

1–6. IEEE.

Alvarez Custodio, M. (2019). Autonomous recharging sys-

tem for drones: Detection and landing on the charging

platform.

Benini, A., Rutherford, M. J., and Valavanis, K. P. (2016).

Real-time, gpu-based pose estimation of a uav for

autonomous takeoff and landing. In 2016 IEEE In-

ternational Conference on Robotics and Automation

(ICRA), pages 3463–3470. IEEE.

Bloesch, M., Burri, M., Omari, S., Hutter, M., and Sieg-

wart, R. (2017). Iterated extended kalman ﬁlter

based visual-inertial odometry using direct photomet-

ric feedback. The International Journal of Robotics

Research, 36(10):1053–1072.

Carlos, B. B., Sartor, T., Zanelli, A., Frison, G., Burgard,

W., Diehl, M., and Oriolo, G. (2020). An efﬁcient

real-time nmpc for quadrotor position control under

communication time-delay. In 2020 16th Interna-

tional Conference on Control, Automation, Robotics

and Vision (ICARCV), pages 982–989. IEEE.

Cocchioni, F., Mancini, A., and Longhi, S. (2014). Au-

tonomous navigation, landing and recharge of a

quadrotor using artiﬁcial vision. In 2014 international

conference on unmanned aircraft systems (ICUAS),

pages 418–429. IEEE.

Daudelin, J., Jing, G., Tosun, T., Yim, M., Kress-Gazit, H.,

and Campbell, M. (2018). An integrated system for

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

110

perception-driven autonomy with modular robots. Sci-

ence Robotics, 3(23).

Garrido-Jurado, S., Munoz-Salinas, R., Madrid-Cuevas,

F. J., and Medina-Carnicer, R. (2016). Generation of

ﬁducial marker dictionaries using mixed integer linear

programming. Pattern Recognition, 51:481–491.

Hornung, A., Wurm, K. M., Bennewitz, M., Stachniss,

C., and Burgard, W. (2013). OctoMap: An efﬁ-

cient probabilistic 3D mapping framework based on

octrees. Autonomous Robots. Software available at

http://octomap.github.com.

Jain, K. P. and Mueller, M. W. (2020). Flying batter-

ies: In-ﬂight battery switching to increase multirotor

ﬂight time. In 2020 IEEE International Conference on

Robotics and Automation (ICRA), pages 3510–3516.

IEEE.

Jain, K. P., Park, M., and Mueller, M. W. (2020). Docking

two multirotors in midair using relative vision mea-

surements. arXiv preprint arXiv:2011.05565.

Kamel, M., Burri, M., and Siegwart, R. (2016). Linear

vs Nonlinear MPC for Trajectory Tracking Applied to

Rotary Wing Micro Aerial Vehicles. ArXiv e-prints.

Kamel, M., Stastny, T., Alexis, K., and Siegwart, R. (2017).

Model predictive control for trajectory tracking of un-

manned aerial vehicles using robot operating system.

In Koubaa, A., editor, Robot Operating System (ROS)

The Complete Reference, Volume 2. Springer.

Lee, T., Leok, M., and McClamroch, N. H. (2010). Geo-

metric tracking control of a quadrotor uav on se (3). In

49th IEEE conference on decision and control (CDC),

pages 5420–5425. IEEE.

Lin, Y., Gao, F., Qin, T., Gao, W., Liu, T., Wu, W., Yang,

Z., and Shen, S. (2018). Autonomous aerial naviga-

tion using monocular visual-inertial fusion. Journal

of Field Robotics, 35(1):23–51.

Lin, Y.-C., Cheng, Y.-T., Zhou, T., Ravi, R., Hasheminasab,

S. M., Flatt, J. E., Troy, C., and Habib, A. (2019).

Evaluation of uav lidar for mapping coastal environ-

ments. Remote Sensing, 11(24):2893.

Loianno, G., Spurny, V., Thomas, J., Baca, T., Thakur, D.,

Hert, D., Penicka, R., Krajnik, T., Zhou, A., Cho, A.,

et al. (2018). Localization, grasping, and transporta-

tion of magnetic objects by a team of mavs in chal-

lenging desert-like environments. IEEE Robotics and

Automation Letters, 3(3):1576–1583.

Lutz, P., M

uller, M. G., Maier, M., Stoneman, S., Tomi

c, T.,

von Bargen, I., Schuster, M. J., Steidle, F., Wedler,

A., St

urzl, W., et al. (2020). Ardea—an mav with

skills for future planetary missions. Journal of Field

Robotics, 37(4):515–551.

Narv

aez, E., Ravankar, A. A., Ravankar, A., Emaru, T., and

Kobayashi, Y. (2020). Autonomous vtol-uav dock-

ing system for heterogeneous multirobot team. IEEE

Transactions on Instrumentation and Measurement,

70:1–18.

Narv

aez, E., Ravankar, A. A., Ravankar, A., Kobayashi,

Y., and Emaru, T. (2017). Vision based autonomous

docking of vtol uav using a mobile robot manipula-

tor. In 2017 IEEE/SICE International Symposium on

System Integration (SII), pages 157–163. IEEE.

Peng, K., Du, J., Lu, F., Sun, Q., Dong, Y., Zhou, P., and

Hu, M. (2019). A hybrid genetic algorithm on routing

and scheduling for vehicle-assisted multi-drone parcel

delivery. IEEE Access, 7:49191–49200.

Perez-Grau, F. J., Caballero, F., Merino, L., and Viguria,

A. (2017). Multi-modal mapping and localization

of unmanned aerial robots based on ultra-wideband

and rgb-d sensing. In 2017 IEEE/RSJ International

Conference on Intelligent Robots and Systems (IROS),

pages 3495–3502. IEEE.

Qin, T., Li, P., and Shen, S. (2018). Vins-mono: A robust

and versatile monocular visual-inertial state estimator.

IEEE Transactions on Robotics, 34(4):1004–1020.

Qin, T. and Shen, S. (2018). Online temporal calibration for

monocular visual-inertial systems. In 2018 IEEE/RSJ

International Conference on Intelligent Robots and

Systems (IROS), pages 3662–3669. IEEE.

Rocha, R. and Robinson, S. K. (2020). Toward autonomous

in-ﬂight docking of unmanned multi-rotor aerial vehi-

cles. In AIAA Scitech 2020 Forum, page 1486.

Romero-Ramire, F. J., Munoz-Salinas, R., and Medina-

Carnicer, R. (2019). Fractal markers: a new approach

for long-range marker pose estimation under occlu-

sion. IEEE Access, 7:169908–169919.

Romero-Ramirez, F. J., Mu

noz-Salinas, R., and Medina-

Carnicer, R. (2018). Speeded up detection of squared

ﬁducial markers. Image and vision Computing,

76:38–47.

Sa, I., Kamel, M., Khanna, R., Popovi

c, M., Nieto, J., and

Siegwart, R. (2018). Dynamic system identiﬁcation,

and control for a cost-effective and open-source multi-

rotor mav. In Field and Service Robotics, pages 605–

620. Springer.

Sani, M. F. and Karimian, G. (2017). Automatic naviga-

tion and landing of an indoor ar. drone quadrotor us-

ing aruco marker and inertial sensors. In 2017 interna-

tional conference on computer and drone applications

(IConDA), pages 102–107. IEEE.

Tzoumanikas, D., Li, W., Grimm, M., Zhang, K., Kovac,

M., and Leutenegger, S. (2019). Fully autonomous

micro air vehicle ﬂight and landing on a moving target

using visual–inertial estimation and model-predictive

control. Journal of Field Robotics, 36(1):49–77.

Ullah, N., Mehmood, Y., Aslam, J., Ali, A., and Iqbal, J.

(2021). Uavs-ugv leader follower formation using

adaptive non-singular terminal super twisting sliding

mode control. IEEE Access, 9:74385–74405.

Wasim, M., Ullah, M., and Iqbal, J. (2019). Gain-scheduled

proportional integral derivative control of taxi model

of unmanned aerial vehicles. Revue Roumaine des

Sciences Techniques-Serie Electrotechnique et Ener-

getique, 64(1):75–80.

Yang, S., Scherer, S. A., Schauwecker, K., and Zell, A.

(2014). Autonomous landing of mavs on an arbitrarily

textured landing site using onboard monocular vision.

Journal of Intelligent & Robotic Systems, 74(1):27–

43.

Yu, C., Cai, J., and Chen, Q. (2017). Multi-resolution visual

ﬁducial and assistant navigation system for unmanned

An Unmanned Aerial Carrier and Anchoring Mechanism for Transporting Companion UAVs

111

aerial vehicle landing. Aerospace Science and Tech-

nology, 67:249–256.

Yu, K., Budhiraja, A. K., and Tokekar, P. (2018). Al-

gorithms for routing of unmanned aerial vehicles

with mobile recharging stations. In 2018 IEEE In-

ternational Conference on Robotics and Automation

(ICRA), pages 5720–5725. IEEE.

Zhang, J. and Singh, S. (2014). Loam: Lidar odometry

and mapping in real-time. In Robotics: Science and

Systems, volume 2.

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

112