IEEE FORMAT HOMEWORK - 3 1

Distributed Guiding Vector Field Control for

Integrated Multi-UAV–USV Systems
Md. Rakibul Islam

[5], [6] explores the potential of integrating UAV-USV tasks,

Abstract— This article discusses a novel control mechanism- a but it is limited to a single UAV and a single USV. Controlling
distributed guiding-vector field (DGVF) controller found in the multiple UAVs and USVs together is much harder because of
literature, which guides unmanned aerial vehicles (UAVs) and
speed differences, disturbances, and uncertainties.
unmanned surface vehicles (USVs) to their prescribed routes
efficiently compared to the existing techniques. The control Coordination of multiple UAVs and USVs introduces more
method is two-fold, where the upper tier handles the routing complexity due to these uncertainties and disturbances. Some
strategy and systematizes their formation on the trajectory, studies explored integrated approaches using sophisticated
whereas the lower tier ensures that the UAVs and USVs adhere to control techniques, such as adaptive guiding strategies [7],
the instructions from the upper level. Significantly, unlike existing leader–follower formations [8], and predictive control [9]. Such
systems, which require numerous states from the vehicles, the
studies, however, are still theoretical in kind, relying on
approach requires only one state of information (scalar virtual
coordinate) from the vehicles, making the interconnection Euclidean distance, which necessitates extensive processing
lightweight and reducing computational expenses. The approach and communication, and ignoring environmental disruptions as
also incorporates a mathematical model to assess the convergence well as sensor and actuator limitations.
of the proposed system. To verify the functionality of the proposed The proposed scheme considers every aspect and presents a
system in practice, they conducted an experiment employing three distributed GVF (DGVF) controller with a two-fold
M-100 UAVs, two HUSTER-16 USVs, a HUSTER-12C USV, and
hierarchical structure for heterogeneous cross-domain
Wi-Fi as a communication link.
unmanned systems (CDUSs). While the lower-level monitors
Index Terms— Distributed guiding-vector field (DGVF, and guarantees the top tier's orders are carried out, the higher
unmanned aerial vehicle (UAV), unmanned surface vehicle (USV) tier deals with formation coordination and maneuver strategy.
This method lowers communication, sensor, and computing
costs by having cars share only virtual coordinates. An actual
I. INTRODUCTION experiment with three M-100 UAVs, two HUSTER-16 USVs,

C OOPERATIVE flight of unmanned aerial vehicles a HUSTER-12C USV, and a WiFi connection is then used to
(UAVs) and unmanned surface vehicles (USVs) has validate the schemes.
garnered significant interest due to their low operating costs,
extended operational range, resilience, and accessibility in II. PROPOSED TECHNIQUE
remote settings [1], [2]. The research efforts were initially The author [10] proposes that virtual coordinates integration
focused on either UAVs or USVs separately, not in with Guiding Vector Field (GVF) and adoption of a hierarchical
combination. Many research works in this arena concentrated control strategy may lower the computation need,
on the coordination among multiple UAVs in the case of communication latency, and allow multiple CDUSs to function
monitoring and surveillance applications. The authors in [3] properly in an aerial route.
proposed decentralized coordination with multiple drones for
A. Preliminaries
aerial surveillance. Neural-Swarm2, a technique based on a
deep learning neural network framework, was presented in [4] Guiding Vector Field (GVF) is a vector which directs vehicle
to solve the aeronautical impact that drones have on one another toward a path and moves it along the prescribed path. It is
when flying in proximity. In the case of USVs, extensive defined by the following expression (1) from [10]
studies have been reported as they are very useful in marine 𝑛−1
applications. However, no approach has been taken to integrate phy
the usage of multiple UAVs and USVs in a system together, 𝜒 = × (∇∅1(𝑞),…, ∇∅𝑛−1(𝑞) ) − ∑ 𝑘𝑗 ∅𝑗 (𝑞)∇∅𝑗 (𝑞) (1)
𝑗=1
which may expand the operating zones. While USVs can only
operate on the water's surface, UAVs are constrained by their
batteries. The monitoring region increases substantially from where 𝜒 phy is the guiding vector field, × indicates the cross
2D to 3D when USV fleets are equipped with onboard UAV product, 𝑘𝑗 is the control gain, and ∅𝑗 represents the deviation
swarms. This allows for more sophisticated applications, distance from the defined path. If 𝜒 phy = 𝟎 at certain singular
including maritime-air surveillance, detection, rescue points, then the vehicle stands still or behaves unpredictably. To
operations, and combating illicit trade. Some recent research counter this, virtual coordinate 𝝎 is added, creating a higher-
IEEE FORMAT HOMEWORK - 3 2

dimensional GVF that enables working globally and vehicles The top plot of Fig. 1 from [10] depicts that the path-
starting from anywhere follow the route described [10]. The following error goes down to zero eventually, the middle plot
authors conder a CDUS consisting of 𝑛 USVs represented by indicates the spacing between the nearby vehicles stabilizes
𝑉1 = {1, 2, … , 𝑛} and 𝑚 UAVs represented by 𝑉2 = {𝑛 + over time, and the bottom plot shows the rate of change in
1, 𝑛 + 2, … , 𝑛 + 𝑚}. The USVs path is modeled in 2D, and vehicles' virtual position reach a constant value indicating
UAVs path is modeled in 3D where the paths can be shifted into smooth movement, proving the efficacy of the proposed
higher dimensions using the virtual coordinate 𝝎. Here, the technique in coordinating multiple unmanned vehicles [10].
challenge of coordination is to align 2D and 3D vehicles to The findings offer compelling proof that in complicated
trajectories and diverse vehicle dynamics, proposed guiding
move along their paths at the right speeds [10]. To ensure the
and coordination startegy permits the seamless, steady, and
vehicles are maintaining the defined formation, communication
coordinated operation of numerous UAVs and USVs. The
is developed to achieve path convergence, path maneuvering, technique demonstrated its potential for use in actual multi-
and utilize virtual coordinates instead of full positions to reduce vehicle missions, including environmental monitoring,
the load from communication and sensors. The proposed model surveillance, and cooperative transport.
includes variables for error correction in path following and the
method combines path following with velocity tracking. The III. CONCLUSION
control mechanism is split into two stages. The first stage,
termed the lower level, handles the vehicle’s own dynamics In this article, a DGVF controller technique for
using actuator commands and calculates the required heterogeneous CDUSs is reviewed, which seems to perform
corrections in speed to reduce path-following errors. These satisfactorily. They introduced two levels of control: the upper
tier manages the routing strategy and organizes the formation
outputs are then fed into the second stage, defined as the upper
of the UAVs and USVs on the trajectory, while the lower tier
level, which uses the Guiding Vector Field (GVF) to generate
makes sure the UAVs and USVs follow the higher level's
commands that guide the vehicle along the paths and maintain commands. Importantly, the approach requires only scalar
formation. The authors [10] went through multiple stages of virtual coordinates from the vehicles, meaning the
validation for their mathematical framework. First, they showed interconnection is lightweight and computational costs are
that the position and velocity errors converge to zero. Second, reduced. Finally, the authors verified the efficacy of the
they demonstrated that error states and virtual coordinates technique through an experiment.
remain confined using Lyapunov analyses. Thus, it is verified
that both lower and upper levels are operating correctly. To REFERENCES
ensure formation keeping between USVs and UAVs, they
further verify that the relative positions and virtual coordinates [1] Y. Chen, D. Chang, and C. Zhang, “Autonomous tracking using a swarm
between vehicles are maintained. of UAVs: A constrained multi-agent reinforcement learning approach,”
IEEE Trans. Veh. Technol., vol. 69, no. 11, pp. 13702–13717, Nov. 2020.
B. Experimental Results [2] B.-B. Hu and H.-T. Zhang, “Bearing-only motional target-surrounding
The researchers [10] conducted a real lake experiment with control for multiple unmanned surface vessels,” IEEE Trans. Ind.
Electron., vol. 69, no. 4, pp. 3988–3997, Apr. 2022.
three M-100 UAVs, two HUSTER-16 USVs, and a HUSTER- [3] H. Huang, A. V. Savkin, and C. Huang, “Decentralized autonomous
12C USV. The vehicles effectively maintain formation and the navigation of a UAV network for road traffic monitoring,” IEEE Trans.
required intervehicle distance while following the designated Aerosp. Electron. Syst., vol. 57, no. 4, pp. 2558–2564, Aug. 2021.
[4] G. Shi, W. Hönig, X. Shi, Y. Yue, and S.-J. Chung, “Neural-Swarm2:
circular and self-intersecting pathways. Planning and control of heterogeneous multirotor swarms using learned
interactions,” IEEE Trans. Robot., vol. 38, no. 2, pp. 1063–1079, Apr.
2022.
[5] Y. Meng, W. Wang, H. Han, and M. Zhang, “A vision/radar/INS
integrated guidance method for shipboard landing,” IEEE Trans. Ind.
Electron., vol. 66, no. 11, pp. 8803–8810, Nov. 2019.
[6] S. Lee et al., “Sliding mode guidance and control for UAV carrier
landing,” IEEE Trans. Aerosp. Electron. Syst., vol. 55, no. 2, pp. 951–
966, Apr. 2019.
[7] H. Su, M. Long, and Z. Zeng, “Controllability of two-time-scale discrete-
time multiagent systems,” IEEE Trans. Cybern., vol. 50, no. 4, pp. 1440–
1449, Apr. 2020.
[8] H.-T. Zhang et al., “Visual navigation and landing control of an
unmanned aerial vehicle on a moving autonomous surface vehicle via
adaptive learning,” IEEE Trans. Neural Netw. Learn. Syst., vol. 32, no.
12, pp. 5345–5355, Dec. 2021.
[9] D. Huang, H. Li, and X. Li, “Formation of generic UAVs-USVs system
under distributed model predictive control scheme,” IEEE Trans. Circuits
Syst. II, Exp. Briefs, vol. 67, no. 12, pp. 3123–3127, Dec. 2020.
[10] B.-B. Hu et al., “Coordinated navigation control of cross-domain
Fig. 1. Path-following errors (top plot), the intervehicle parametric unmanned systems via guiding vector fields,” IEEE Trans. Control Syst.
parameters (mid plot), and the derivative of the virtual coordinates (bottom Technol., vol. 32, no. 2, pp. 550–563, Mar. 2024.
plot), all show changes with time [10].