Review of Tethered Unmanned Aerial Vehicles: Building Versatile and Robust Tethered Multirotor UAV System

Abstract

This paper presents a comprehensive review of tethered unmanned aerial vehicles (UAVs), focusing on their challenges and potential applications across various domains. We analyze the dynamic characteristics of tethered UAV systems and address the unique challenges they present, including complex tether dynamics, impulsive forces, and entanglement risks. Additionally, we explore application-specific challenges in areas such as payload transportation and ground-connected systems. The review also examines existing tethered UAV testbed designs, highlighting their strengths and limitations in both simulation and experimental settings. We discuss advancements in multi-UAV cooperation, ground–air collaboration through tethers, and the integration of retractable tether systems. Moreover, we identify critical future challenges in developing tethered UAV systems, emphasizing the need for robust control strategies and innovative solutions for dynamic and cluttered environments. Finally, the paper provides insights into the future potential of variable-length tethered UAV systems, exploring how these systems can enhance versatility, improve operational safety, and expand the range of feasible applications in industries such as logistics, emergency response, and environmental monitoring.

1. Introduction

Tethers (or cables) are widely used across various fields due to their versatility in solving a range of problems. They are commonly employed in payload transportation, such as cranes in construction and tethered helicopters. Tethered helicopters, in particular, are used in applications requiring flexibility, including military supply transport, search and rescue operations, and wildfire suppression. Compared to alternatives like grapplers or platforms, tethers offer advantages such as accommodating diverse payload shapes and enabling simpler mechanisms. For instance, fixed-wing tethered UAVs (TUAVs) are used to transport goods with energy efficiency [1] and to the remote area. Beyond transportation, cables are also used to maintain stable connections between agents, such as communication beacons [2,3,4,5] and underwater or surface water operations [6,7,8,9]. Tethered multirotor unmanned aerial vehicles (TMUAVs) can serve as temporary communication beacons, enabling emergency communication in remote areas or disaster zones [10]. However, integrating cables with aerial vehicles introduces unique challenges. These include impulsive dynamics caused by cable collisions when transitioning between slack and taut states, complex coupled dynamics between aerial vehicles and cables, and robustness issues due to additional external disturbances. Moreover, the physical constraints of multirotor UAVs further limit the practical applications of tethered multirotor systems.

The recent development of small, affordable UAVs, such as quadrotors and hexarotors, has generated interest in using these platforms with tethers for smaller-scale tasks. These tasks- such as confined environment operations [11,12], missions in residential areas [13], and small-item transportation [14]—were previously unsuitable for tethered helicopters. However, TMUAVs face distinct challenges beyond those of tethered helicopters, including limited power and payload capacities, as well as increased vulnerability to external disturbances. Early research on TMUAVs [15,16,17,18] explored their use in various settings, with many of these studies adapting tethered helicopter platforms for smaller UAV applications. Additionally, several early works simplified the system by modeling the tether as a rigid rod [7,19,20], thereby neglecting the tether’s flexibility. This assumption restricted the system’s dynamics by omitting cable collision effects.

To address these limitations, more recent studies have focused on capturing the unique dynamics of TMUAV systems. The incorporation of flexible tether dynamics [21] has improved the modeling of complex interactions in payload transportation. Geometric controllers have been developed to enhance the control of TMUAV systems [22], while optimal control strategies for TMUAVs with flexible cables have been introduced [23]. Given the limited capacity of individual UAVs, multi-vehicle approaches have been explored for slung payload transportation [24,25,26,27]. Additionally, heterogeneous multi-agent tethered UAV systems [28,29] have been proposed to enable more versatile operations. Ground-anchored TMUAV systems have also been investigated [30,31,32], where a motorized ground station actively controls the TMUAV and maintains tether tension throughout the operation. These advancements contribute to overcoming the constraints of TMUAVs, enabling broader applications and improved system performance.

This paper explores recent advancements in TMUAV systems across various domains, with a particular focus on system modeling and control architecture. As TMUAVs operate in environments where traditional untethered UAVs face significant limitations, understanding their unique characteristics is essential for optimizing performance and expanding their applicability. Unlike conventional multirotor UAVs, TMUAVs experience tether-induced forces that introduce additional complexities, including impulsive dynamics from slack-to-taut transitions, increased susceptibility to external disturbances, and maneuverability constraints. Through an in-depth review of these developments, this paper provides valuable insights into the effective utilization of TMUAVs in future technologies. In addition to highlighting recent progress in TMUAV research, we identify key challenges and research gaps that must be addressed to further enhance their capabilities and practical deployment. The key advantages and challenges for tethered UAV system are summarized in

Table 1. Advantages and challenges of tethered UAV systems.

Advantages	Challenges
No need to land on rough terrain	Impulsive cable dynamics
Power supply for extended missions	Tether weight
Guided landing	Tether entanglement
Undisturbed communication	Higher hardware complexity
Collaboration with ground robot	Higher disturbance
Navigation in enclosed spaces	Obstacle avoidance

.

In recent years, there have been few review and survey papers on tethered UAVs. For example, Ref. [33] examined the potential of networked tethered aerial platforms for various applications. A systematic review presented in [34] focused on categorizing and summarizing recently published solutions for TUAV systems. In contrast to these works, this paper focuses specifically on the challenges associated with controller design for tethered multirotor UAVs. We highlight key issues related to tether modeling and the design of cooperative control architectures. Additionally, we present recent advancements in experimental testbeds developed to address these challenges and validate proposed solutions. Finally, we outline future research directions in the area of variable-length TMUAVs, introducing recent developments and identifying open research questions that remain to be explored.

The remainder of this paper is structured as follows: Section 2 provides an overview of TMUAV applications, highlighting key use cases along with their strengths and limitations. Section 3 discusses the challenges in modeling TMUAV system and designing robust control framework. Section 4 presents real-world TMUAV testbeds, where different control strategies and architectures have been implemented. Specifically, we introduce works on single TMUAV, multi-agent systems, and heterogeneous cooperative systems. Section 5 outlines key future potentials for varying length TMUAV applications across different domains. Specifically, we focus on exploring how MUAV with a retractable tether system can address the real-world problems. Finally, concluding remarks and future research considerations are presented in Section 6.

2. Applications of TMUAVs

This section identifies current key applications of TMUAVs across various domains. Unlike conventional tethered aerial vehicles, such as helicopter systems primarily used for payload transportation, smaller TMUAVs offer greater versatility at a lower cost. Their reduced maintenance and operation requirements compared to helicopters make them suitable for applications such as temporary communication beacons in disaster zones providing local network coverage. Moreover, traditional tethered helicopters are not practical in dense environments, such as residential areas and cluttered indoor spaces, where smaller TMUAVs can navigate effectively. The remainder of this section highlights key TMUAV applications and examines the critical technologies that enable each use case.

2.1. Payload Transportation

Payload transportation is one of the most common applications of tethered aerial vehicles, particularly in scenarios where traditional logistics methods may not be feasible or efficient. Tethered UAVs offer a significant advantage in navigating challenging environments, such as rugged terrains, dense forests, urban canyons, and disaster zones. One of the primary benefits of using tethers for payload transportation is the ability to hover and deploy payloads without the need for the UAV to land. This capability is especially useful in situations where landing could be dangerous or impossible, such as on uneven ground, in water, or in areas with debris or obstacles. The tether provides not only stability but also the ability to raise and lower the payload safely while maintaining a controlled position above the deployment site. For example, in search and rescue operations, a tethered UAV can lower medical supplies or communication devices into hard-to-reach locations without risking the safety of rescue personnel.

Adaptive controllers have been utilized to manage TMUAVs in tethered payload transportation tasks due to their ability to handle uncertainties and dynamic changes in system behavior. These controllers adjust control parameters in real-time, allowing the system to maintain stability and performance despite varying conditions. In [35], adaptive controller was developed to account for shifts in the center of gravity during payload transportation. These shifts occur as the payload moves or swings, altering the system’s dynamics and requiring the control strategy to adapt continuously to maintain balance and precision. Additionally, swing-free trajectory generation method based on dynamic programming was proposed to minimize the effect of swing during aggressive maneuver. The proposed framework for agile maneuver was validated through simulation and experiment. Ref. [36] proposed an adaptive control for a quadrotor with a suspended load by an elastic rope, which is modeled as a single mass-spring-damper system. The performance of trajectory following was validated through numerical simulation.

A geometric controller for cable-suspended payloads was proposed in [37,38]. The geometric controller operates in a coordinate-free framework, allowing for more efficient control of the TMUAV’s movements. Additionally, the system used a hybrid cable model to capture the complex interactions between the UAV, the payload, and the tether. In [22], the flexible tether was modeled using a series of links, which considers not only the translational movements of the payload but also the rotational and oscillatory dynamics introduced by the flexible tether. The effectiveness of this geometric controller was experimentally validated in [39]. The experimental results showed that the geometric controller significantly reduced oscillations and improved the overall stability of the system. Furthermore, an adaptive geometric controller was proposed in [40] to decouple attitude dynamics.

The choice of a geometric controller offers several advantages for payload transportation TMUAV systems, including robustness under dynamic disturbances and with varying payload conditions. Specifically, Refs. [37,38] showed that TMUAV system can be expressed by differentially flat hybrid system that can handle both slack and taut conditions of the cable. This allows more robust operation by addressing the cable collision due to external disturbance. Ref. [41] provides proofs for the almost global exponential stability of both quadrotor attitude controlled flight mode and load attitude controlled flight mode, as well as the exponential stability of load position and cable length controlled flight mode. Ref. [22] presents a rigorous Lyapunov stability analysis for their geometric controller and ensures robustness against uncertainties in both rotational and translational dynamics of the flexible cable TMUAV model via an added nonlinear integral control term. This is shown to be useful in especially aggressive maneuvers which cause significant cable deformation.

An adaptive controller tailored for a VTOL (Vertical Take-Off and Landing) tethered payload transportation system was introduced in [42]. This system demonstrated an improved performance in complex operational scenarios by combining the adaptability of the controller with the versatility of a VTOL platform. The controller dynamically adjusted the UAV’s flight path to minimize load swing and maintain a stable payload orientation, even when external disturbances, such as wind gusts or abrupt changes in direction, occurred. These control strategies offer robust solutions for managing the intricate dynamics of TMUAV systems.

Trajectory generation for TMUAVs with suspended payloads requires careful consideration of additional constraints to effectively manage payload dynamics and minimize load swing. This is critical in ensuring both the safety and efficiency of operations, particularly in scenarios involving precise payload placement or navigation through confined spaces. A method for trajectory generation under the assumption of minimal load swing was proposed in [35]. This approach involves generating smooth trajectories that reduce oscillatory behavior by aligning the motion of the UAV with the dynamics of the suspended payload. By maintaining minimal swing, this method improves payload stability and reduces the risk of collision with nearby obstacles.

In [43], trajectory generation approach was introduced using a coordinate-free dynamic model for a differentially flat hybrid system of TMUAVs. This method does not rely on a fixed coordinate system, allowing for more flexible and efficient path planning. A mixed-integer quadratic programming (MIQP) approach to trajectory generation was developed in [44]. This method optimizes complex motion paths by integrating tethered constraints directly into the planning algorithm. The MIQP framework allows for precise management of state and input constraints, enabling the TMUAV to execute complex maneuvers while avoiding collisions and ensuring the payload remains stable.

To further address payload swing, a trajectory specifically designed to suppress load oscillation was proposed in [45]. This method enhances stability during transport, making it suitable for sensitive payloads or operations in environments where precision is crucial. In [46], a trajectory generation technique for a quadrotor lifting a payload was explored, demonstrating practical applications of trajectory planning in real-world scenarios. This technique involved simulating the interaction between the quadrotor and the suspended payload under varying flight conditions. These studies contribute to a deeper understanding of how to generate effective trajectories for TMUAVs with suspended payloads, offering solutions that enhance stability, safety, and operational efficiency in a wide range of applications.

While TMUAVs offer enhanced versatility, their payload capacity is limited compared to conventional aerial vehicles. To overcome this limitation, multiple TMUAVs can operate cooperatively to transport heavy or irregularly shaped payloads, allowing the system to exceed the load capacity of a single UAV while maintaining stability. For example, a cooperative payload transportation system using two TMUAVs for a bar-shaped payload was demonstrated in [47]. This approach treats the payload and TMUAVs as a single integrated system, utilizing multi-robot formation kinematics. An adaptive dynamic compensator manages the complex dynamics of the tethered system and payload, ensuring accurate tracking, precise payload positioning, and stable payload orientation even in the presence of external disturbances.

In [25], a multi-UAV collaborative transportation method with obstacle avoidance was introduced using a 2D model. Additionally, a comprehensive framework for collaborative TMUAV systems with multiple UAVs was developed in [26]. An outdoor cooperative payload transportation approach using TMUAVs was proposed in [48], demonstrating practical implementation in real-world environments. A reinforcement learning-based low-level controller for two quadrotors with a slung payload was presented in [49], showcasing the potential of machine learning techniques in enhancing system performance. Furthermore, a geometric controller specifically designed for cooperative multi-TMUAV transportation systems was proposed in [50], highlighting advancements in control strategies for complex multi-agent scenarios.

2.2. Disaster Management

Disasters such as hurricanes and earthquakes can devastate vital communications infrastructure for extended periods, necessitating effective emergency disaster management. During these critical times, TMUAV systems equipped with onboard antennas and cell towers can rapidly restore communication services. These systems support recovery efforts for first responders and facilitate a quicker return to normalcy for affected communities.

In [10], a flying network for emergency situations using tethered multicopters is proposed, demonstrating how TMUAVs can establish temporary communication networks. TMUAV can also be used to provide continuous supply the data connection with ground terminals [3]. A robotic car equipped with essential communication equipment, including a TMUAV, was introduced in [5], highlighting a mobile and adaptable approach to reestablishing communication infrastructure. Additionally, a position estimation framework developed in [51] offers visual support for teleoperated construction machines in disaster scenarios, enhancing safety and operational efficiency. The application of tethered UAVs for providing wireless communication during disasters is further explored in [8].

The Oxpecker system, introduced in [12], can monitor structural health in disaster scenarios, adding another layer of utility to TMUAV deployments. A key advantage of using TMUAVs in emergency scenarios is that FAA regulations are generally more permissive for tethered UAVs compared to untethered flights. This regulatory flexibility enables quicker deployment by companies and citizens, as there is often no need for special permits, allowing for a more agile response to emergencies.

Currently, most search and rescue missions are carried out on foot or with manned vehicles, requiring labor and risk to human operators. The search and rescue task can be made safer through the use of UAVs while increasing mission speed and versatility. Applying existing TMUAV capabilities to a search and rescue mission may allow autonomous location and even transportation of lost or injured individuals in rugged terrain otherwise difficult or unsafe to reach [52]. In [53], the network performance of tethered UAV-assisted intelligent edge computing was evaluated. Cooperative TMUAV payload transportation may be applied for mountain rescue missions, or collaborative TMUAV–USV systems may be used to rescue individuals lost in bodies of water [54].

2.3. Data Collection

TMUAVs are increasingly being utilized in data collection missions. Specifically, they are used in marine environments for applications such as data collection, underwater operations, and surveillance. Early work in this domain, such as [6], proposed the use of a TMUAV attached to a marine vessel for visual observation and identification of oil slicks. The system used a simplified tether model with only 10 finite elements, enabling real-time computing.

In [7], a floating buoy was attached to a TMUAV and dragged across the water surface via the tether. The researchers modeled the water medium, buoy, and tether separately and integrated them using the Euler–Lagrange formulation. To maintain a taut tether and ensure consistent buoy contact with the water surface, a precision motion control system for surge velocity was developed, along with attainable setpoints and constraints. This approach could enhance unmanned search and rescue capabilities and support offshore wind turbine or pipeline repairs.

Ref. [9] introduced a system where a TMUAV is connected to a self-propelled unmanned surface vehicle (USV) in rough sea conditions, with the tether in a slack state. This concept features a motorized reel and sensors on the USV, using a catenary curve estimation to approximate the tether position based on tether angle, length, and tension. The motor ensures the tether remains in a semi-slack state, accommodating the dynamic nature of the marine environment.

3. Challenges of TMUAVs

This section examines the unique challenges involved in designing and controlling TMUAV systems. A key focus is on the modeling complexities introduced by tethers, which exhibit complex and impulsive dynamics, particularly during transitions between slack and taut states. Various cable modeling methods are introduced, along with control design approaches to manage these challenging dynamics effectively. Additionally, we explore the development of cooperative frameworks for multi-agent tethered systems, addressing the specific challenges of coordinating air–air, ground–air, and underwater interactions. These cooperative strategies are essential for expanding the versatility and operational efficiency of TMUAV systems in diverse environments and applications.

3.1. Dynamics of Tether

One of the primary challenges in TMUAV systems lies in managing the complex dynamics and geometry of the tether. Tether oscillations are difficult to model accurately, and the discrete transitions between slack and taut states introduce unwanted impulsive forces into the system. Entanglement of the tether is another critical concern, especially in environments with obstacles or when operating multiple UAVs simultaneously. Additionally, the dynamics of the tether are time-varying, as both the tether length and payload weight can change throughout different mission stages. This variability demands precise real-time computation to maintain stability and control. However, this real-time computational requirement presents a significant challenge for small TMUAVs due to their limited onboard processing capacity. Figure 1 illustrates two different methods of modeling cable in a TMUAV system.

Tether dynamics can be simplified by modeling the tether as a rigid rod, as proposed in [15]. This assumption eliminates the impulsive dynamics caused by the discrete state transitions between slack and taut conditions, thereby simplifying the system’s equations of motion. In [51], an estimation method for non-taut cables is introduced, implementing a tether winding mechanism based on a predictable catenary curve derived from tether angle and tension. This approach provides a practical solution for managing slack in the tether while maintaining system stability. An alternative approach to approximate tether dynamics involves modeling the tether as a series of rigid links.

If the flexibility of the cable is neglected, the quadrotor and load position can be described by

$$\begin{array}{cc}\hfill x_Q& =x_L-lq,\hfill \end{array}$$

(1)

where $x_Q\in {\mathbb{R}}^3$ is the position vector of the center of mass of the quadrotor in the inertial frame shown in Figure 1, $x_L$ is the position vector of the load in the inertial frame, l is the length of the cable, and q is the unit vector from quadrotor to the load. Then, the equation of motion of TMUAV with non-zero tension is given by

$$\begin{array}{cc}\hfill {\dot{x}}_L& =v_L,\hfill \end{array}$$

(2)

$$\begin{array}{cc}\hfill (m_Q+m_L)({\dot{v}}_L+g{e}_z)& =(q·fR{e}_3-m_{Q}l(\dot{q}·\dot{q})q,\hfill \end{array}$$

(3)

$$\begin{array}{cc}\hfill \dot{q}& =w\times q,\hfill \end{array}$$

(4)

$$\begin{array}{cc}\hfill m_{Q}l\dot{\omega }& =-q\times fR{e}_z,\hfill \end{array}$$

(5)

$$\begin{array}{cc}\hfill \dot{R}& =R\widehat{\Omega },\hfill \end{array}$$

(6)

$$\begin{array}{cc}\hfill J_Q\dot{\Omega }+\Omega \times J_Q\Omega & =M,\hfill \end{array}$$

(7)

where $v_L$ is the velocity vector of the quadrotor, $m_Q$ and $m_L$ are masses of quadrotor and load, respectively, g is a gravitational constant, f is the magnitude of the thrust for the quadrotor, R is the rotation matrix of the quadrotor from body-fixed frame to the inertial frame, $\omega$ is the angular velocity of the load, $\Omega$ is the angular velocity of the quadrotor in the body-fixed frame, hat map $\widehat{·}$ is defined such that $\widehat{x}y=x\times y$, $J_Q$ is the inertia matrix of the quadrotor, and M is the moment vector for the quadrotor. Furthermore, if the tension in the cable becomes zero, the equation of motion becomes

$$\begin{array}{cc}\hfill {\dot{x}}_L& =v_L,\hfill \end{array}$$

(8)

$$\begin{array}{cc}\hfill m_L({\dot{v}}_L+g{e}_z)& =0,\hfill \end{array}$$

(9)

$$\begin{array}{cc}\hfill {\dot{x}}_Q& =v_Q,\hfill \end{array}$$

(10)

$$\begin{array}{cc}\hfill m_Q({\dot{v}}_Q+g{e}_z)& =fR{e}_z,\hfill \end{array}$$

(11)

$$\begin{array}{cc}\hfill \dot{R}& =R\widehat{\Omega },\hfill \end{array}$$

(12)

$$\begin{array}{cc}\hfill J_Q\dot{\Omega }+\Omega \times J_Q\Omega & =M.\hfill \end{array}$$

(13)

The differentially flat hybrid hybrid model can be used to describe these two different modes of TMUAV dynamics. Details of this modeling can be found in [38].

When the cable is modeled as a serially connected links [22], Lagrangian methods could be used to model the system. The kinetic energy and potential energy are given by

$$\begin{array}{cc}\hfill T=& {\displaystyle \frac{1}{2}}\left(m+\sum _{i=1}^n{m}_i\right){\parallel \dot{x}\parallel }^2+\dot{x}\sum _{i=1}^n\left(\sum _{j=i}^n{m}_j{l}_i\right){\dot{q}}_i\hfill \\ \hfill & +{\displaystyle \frac{1}{2}}\sum _{i=1,j=1}^n\left[\left(\sum _{k=maxi,j}^n{m}_k\right)l_i{l}_j\right]{\dot{q}}_i{\dot{q}}_j+{\displaystyle \frac{1}{2}}{\Omega }^{\mathrm{T}}J\Omega ,\hfill \end{array}$$

(14)

$$\begin{array}{cc}\hfill V& =-\sum _{i=1}^n\sum _{j=1}^n{m}_{j}g{l}_i{e}_3{q}_i-\left(m+\sum _{i=1}^n{m}_i\right)g{e}_{3}x,\hfill \end{array}$$

(15)

where n is the number of links, $m_i$, $l_i$, and $q_i$, $i=1,\dots ,n$, be the mass of, the length of, and the unit vector representing the i-th link, respectively, x is the position of the quadrotor, m is the mass of the quadrotor, J is the inertia matrix of the quadrotor, and $\Omega$ is the angular velocity of the quadrotor. This method allows for adjustable computational complexity by varying the number of links. The technique is further refined in [21], where the controller stacks the links vertically to maintain a balanced payload. To prevent entanglement with obstacles, Ref. [24] employs a similar strategy by dividing the tether into multiple rigid links, enhancing maneuverability in complex environments.

A flexible cable model proposed in [23] treats the cable as a linearly elastic structure influenced by gravity and aerodynamic drag, offering a realistic representation of tether behavior in dynamic conditions. The elongation of the cable is modeled as

$$\begin{array}{cc}\hfill \epsilon (s,t)& ={\displaystyle \frac{\parallel {\mathbf{r}}_s(s,t)ds\parallel -ds}{ds}}=\parallel {\mathbf{r}}_s(s,t)\parallel -1,\hfill \end{array}$$

(16)

where $ds$ is an infinitesimal cable length and $\parallel r_s(s,t)\parallel$ is the actual cable differential element length. Then, the internal force that is tangent to the cable is modeled as

$$\begin{array}{c}\hfill \mathbf{n}(s,t)=n(s,t){\displaystyle \frac{{\mathbf{r}}_s(s,t)}{\parallel {\mathbf{r}}_s(s,t)\parallel }},\end{array}$$

(17)

where $n(s,t)$ is the cable’s tension at $(s,t)$. A nonlinear model predictive controller based on reduced-order modeling is then used to optimally track trajectories, balancing system complexity with control performance. This combination of advanced modeling and control techniques enables the TMUAV system to perform precise maneuvers in challenging environments, demonstrating the potential of flexible cable models in enhancing both safety and efficiency in tethered UAV operations.

Table 2. Advantages and Limitations of different tether modeling methods.

Model Type	Advantages	Limitations
Rigid rod [7,14,15,18,19,20,50,55]	Simple dynamics Taught tether	Unrealistic Cannot describe slack tether
Hybrid model [21,37,38,43,44]	Slack and taught tether models Differentially flat hybrid system	Modeling error from approximation Delay in slack-taut transition
Series of links [22,56]	Slack tether behavior More realistic modeling Number of links can be increased for higher fidelity	Modeling error from approximation More computationally demanding
Flexible cable [23,27]	Accurate modeling of elastic tether	Computationally demanding

summarizes characteristics of different tether modeling strategies with key literature. In choosing a tether modeling approach, the tradeoff between modeling accuracy and computational simplicity must be considered for the specific system. For this reason, the choice of tether modeling approach is highly individual, and the simpler approaches are still commonly used in recent papers.

Tether entanglement poses a significant limitation for TMUAV systems operating in confined spaces or around obstacles. Definitions of entanglement, which can be effectively used to qualify the entanglement state of a tethered robot in diverse situations is provided in [57]. To address this challenge, a tether-aware path planning algorithm was introduced in [58]. This approach uses a hierarchical framework for exploration planning within a 3D cavity, enabling the TMUAV to navigate complex environments while minimizing the risk of entanglement. Modeling underground or enclosed systems as tethered systems can also benefit untethered multi-agent systems. This modeling approach ensures consistent communication pathways between agents, which is particularly valuable in environments where communication signals are limited. By simulating tether constraints, the system maintains reliable connections, enhancing mission effectiveness.

A more advanced solution to tether entanglement involving multiple agents is presented in [27]. NEPTUNE system offers a robust approach to trajectory planning for multiple TMUAVs connected to ground stations within obstacle-laden environments. The system employs a tether-aware homotopy model to filter out trajectories that could lead to entanglement. Additionally, a back-end trajectory optimizer refines path planning, ensuring safe and efficient movement of TMUAVs through complex spaces.

3.2. Cooperative TMUAV Mission

Systems employing smaller UAVs face unique challenges in exchange for their maneuverability, low cost, and lightweight design. These UAVs are typically limited in payload capacity, often requiring multiple agents to work collaboratively for payload transportation tasks. Their reduced size also restricts the onboard integration of high-performance sensors, cameras, and communication hardware, necessitating more efficient and simplified control and communication strategies. To address these limitations, Ref. [48] proposed a system using three TMUAVs to cooperatively transport a payload through a self-triggered path planning and coordinated control framework. This approach effectively mitigates bandwidth constraints in multi-agent systems by minimizing unnecessary communication.

Ref. [37] applied geometric control techniques to a system of multiple quadrotors transporting a suspended payload. Numerical simulations demonstrated that the quadrotors could cooperatively ensure a point-mass load followed a desired trajectory with high accuracy. Furthermore, Ref. [21] presented a system where four TMUAVs were used to transport a rigid-body payload, showcasing the practical implementation of geometric control and cooperative strategies in a multi-agent tethered system. These studies collectively highlight the potential of coordinated multi-UAV frameworks in overcoming the inherent limitations of small UAV platforms.

Cooperative TMUAV missions have expanded to include heterogeneous systems involving ground robots and maritime surface vehicles. A key challenge in these systems lies in coordinating control between TMUAVs and ground or surface agents, each with distinct hardware capabilities and operational constraints. The goal is to enable efficient collaboration that capitalizes on the strengths of each platform without overburdening any single agent. Three representative collaborative systems with TMUAV are shown in Figure 2.

A notable example of such cooperation is presented in [59], where a UAV–UGV (Unmanned Ground Vehicle) team is used for cliff-climbing missions. In this setup, the UAV wraps a tether around terrain features to create an anchor, aiding the UGV in ascending steep slopes. The UAV functions as a sensor platform and navigation aid, while the UGV provides power and houses data collection hardware. This configuration is particularly effective for extended rover-type missions in harsh or inaccessible environments.

The collaborative approaches enhances overall system efficiency by combining the UAV’s high vantage point with the UGV’s ground stability. The UAV contributes to obstacle detection and navigation, while the UGV handles ground-level tasks such as transporting heavy equipment or collecting environmental data. By intelligently coordinating the roles of both agents, the system significantly reduces energy consumption and operational time compared to traditional single-agent systems, which might struggle with similar tasks. The method demonstrates a balanced synergy where each agent’s weaknesses are mitigated by the other’s strengths, offering a versatile and adaptive solution for complex environments.

In the maritime domain, UAV–USV (Unmanned Surface Vehicle) collaboration has been explored through various implementations. In [9], a winching mechanism is used to control a TMUAV tethered to a USV, enhancing control precision and enabling stable operations on water. A UAV–USV platform was also presented in [60], featuring a USV deck equipped with sensors, allowing a quadrotor UAV to take off and land safely. Similarly, a UAV–USV system for maritime parallel search missions was developed in [61], demonstrating coordinated search patterns between the two agents. Furthermore, ref. [62] proposed the Logic Virtual Ship–Logic Virtual Aircraft (LVS-LVA) guidance principle for UAV–USV coordination, enabling the generation of reference heading signals to guide the UAV based on the USV’s motion.

Incorporating a tethered connection in these systems offers several advantages, including extended mission duration through continuous power supply, safer takeoff and landing on moving platforms, and increased communication bandwidth between agents. These benefits make tethered UAV integration a promising direction for enhancing the performance and reliability of heterogeneous cooperative systems.

4. TMUAV Testbed

This section introduces various existing testbed environments for TMUAV systems, covering both simulation and experimental setups along with their respective strengths and limitations. The testbeds encompass a range of configurations, including single UAV systems, multiple UAV systems, and heterogeneous collaborative UAV–UGV systems. These testbeds can be categorized into systems where the TMUAV is fixed to a ground station or a self-propelled robot and systems where TMUAVs are used to carry a payload or drag it across a surface, such as in marine applications. Each testbed type offers unique advantages and challenges, contributing to the development and testing of advanced TMUAV control strategies and cooperative frameworks.

4.1. Simulation Testbed

Numerical simulations are commonly conducted once a TMUAV concept is established to evaluate whether the system can function as intended and to compare its effectiveness against existing systems. A MATLAB-based simulation environment was introduced in [15] to test a tethered helicopter landing concept. While this environment allows for quick and simple validation of the proof of concept, it lacks the complexity needed to fully validate the model under more realistic conditions. In [58], a more advanced simulation approach was employed by modeling the system in four different environments within ROS using the SLAM approach with the RTAB-Map algorithm. This method offers a robust simulation framework, allowing for a detailed analysis of the system’s performance in dynamic and complex scenarios.

For the slung payload transportation system, where a quadrotor functions as a PVTOL (Planar Vertical Take-Off and Landing) machine, the simulation was performed using the software-in-the-loop ground station platform AuRoRA (Autonomous Rotorcraft Research and Development) [42]. The platform was connected to a virtual AR.Drone quadrotor. Two simulations were conducted: a hovering positioning task, where the simulated vehicle executed simultaneous upward/downward and forward/backward motions, and a trajectory tracking task, where the vehicle followed an infinity-shaped (∞) path. These simulations provided insights into the vehicle’s control performance and trajectory-following accuracy.

The NEPTUNE system [27] was also tested numerically, where each robot ran its program in ROS using the commercial solver Gurobi. The most significant simulation involved missions with robots navigating through an obstacle-filled environment and back, requiring tether paths to cross and obstacles to be avoided. This was achieved using the Unity game engine with the AGX Dynamics physics plugin, demonstrating the system’s ability to handle complex real-world scenarios.

In [14], a machine learning design was tested using three indoor environments. One of these environments modeled a cafe, while the other two replicated the group’s physical testbeds, enabling a comparative analysis of discrepancies between simulated and real-world experimental data. In [44], polynomial basis functions were used to simulate an obstacle avoidance mission where a payload had to be carried through a window. One scenario involved navigating through a low window where simply passing through slowly was not feasible. Another scenario required the payload to be released through the window as only the payload could fit through the opening. These simulations highlighted the flexibility and adaptability of TMUAV systems in challenging environments.

A reinforcement learning approach was also utilized in [49] for low-level control of quadrotors in multi-agent collaborative systems. The quadrotor dynamics were decomposed into translational and yawing components, with separate collaborative reinforcement learning applied to each part to enable more efficient training. The method demonstrated a reduction in steady-state errors and prevented overly aggressive control inputs, promoting more efficient and stable control. Similarly, in [42], a quadrotor carrying a suspended payload was constrained to the XZ plane to function similarly to a PVTOL machine. This constraint simplified the control model while maintaining the system’s ability to perform complex tasks within the designated plane.

4.2. Experimental Testbed

Several experimental testbeds for TMUAV systems have been developed over the past decades to explore their capabilities and refine control strategies. These testbeds are designed to evaluate performance in real-world scenarios and to validate theoretical models and control algorithms. Specifically, we introduce three main types: single TMUAV systems, multi-TMUAV systems, and heterogeneous multi-agent systems incorporating TMUAVs.

Single TMUAV testbeds focus on evaluating the performance of a single TMUAV in tasks such as payload transportation, stability control, and precise maneuvering in confined spaces. These testbeds often incorporate motion capture systems to accurately measure the UAV’s position and payload dynamics. Researchers use these setups to fine-tune control algorithms that manage the complex dynamics introduced by the tether, such as slack-to-taut transitions and load swing suppression.

Aggressive and precise payload transportation using TMUAV was studied in [44]. A Mixed Integer Quadratic Program (MIQP) employed as a trajectory planner within a hybrid dynamical system model. The system demonstrated its ability to perform complex missions by leveraging tether and payload geometry during swinging motions to navigate through challenging obstacles, such as windows, that would otherwise be impassable.

In [35], an adaptive controller for TMUAVs was tested using AscTech Hummingbird quadrotors with a Vicon motion capture system. The testbed included a maze of obstacles to simulate cluttered environments. The quadrotor maintained a position above the obstacles while the payload hung between them. The system demonstrated the ability to execute agile transportation missions, following the optimal path and using less energy for load displacement.

A slung payload transportation system utilizing machine learning to navigate environments with static obstacles was introduced in [14], where an experimental setup aimed to minimize load oscillation through reinforcement learning. This approach accounted for complex dynamic behaviors, particularly during intricate or aggressive maneuvers, highlighting the potential of advanced control methods to improve the stability and accuracy of TMUAV systems.

Multi-TMUAV testbeds involve multiple UAVs working cooperatively to perform complex tasks that a single UAV could not accomplish alone. These testbeds are used to test formation control, cooperative payload transportation, and path planning in environments with potential obstacles and dynamic elements. The experimental setups often feature coordinated control systems where each UAV adjusts its behavior based on the shared payload’s state and the movements of other UAVs. Multi-TMUAV testbeds are critical for evaluating self-triggered and adaptive control systems that help maintain formation integrity and avoid tether entanglement.

In [48], a self-triggered cooperative path-following control method for TMUAVs was proposed, enabling multiple UAVs to synchronously carry a slung payload. Each drone was connected to the payload by a 14 m tether. The self-triggered control approach allowed the UAVs to autonomously determine when to update their paths based on system states and environmental feedback, optimizing efficiency and reducing unnecessary communication between agents.

An online formation planning method for TMUAVs was introduced in [26], focusing on real-time path adjustments during payload transportation. The hardware experiment involved three TMUAVs collaboratively transporting a payload. The system dynamically adapted the formation of the UAVs to maintain stability and control, demonstrating the feasibility of using such systems in scenarios requiring continuous path reconfiguration.

An adaptive control approach for cooperatively transporting a bar-shaped payload was presented in [47]. This study employed experiments to model the system dynamics using two quadrotors connected to the payload by flexible cables. The adaptive controller adjusted the UAVs’ behavior in response to disturbances and payload dynamics. In one test scenario, foam plates were attached to the quadrotors to increase drag, enabling the researchers to evaluate the controller’s performance under higher wind disturbance conditions. This setup simulated real-world conditions where external aerodynamic forces could impact the stability of the system.

In [27], a system with three TMUAVs was tested in a scenario simulating warehouse sorting tasks. The UAVs followed crossing paths to replicate complex sorting operations. Although no physical obstacles were introduced during testing, the UAVs demonstrated precise maneuvering to avoid tether entanglement. The test highlighted the potential of the system for use in dynamic and cluttered environments, where maintaining safe and efficient trajectories is critical to operational success.

Heterogeneous multi-agent systems with TMUAVs combine aerial vehicles with ground robots (UGVs) or surface vehicles (USVs) to create versatile and collaborative platforms. These testbeds test scenarios such as search and rescue, environmental monitoring, and logistics operations. The UAVs can provide aerial support for mapping, obstacle detection, or payload transportation, while the ground or surface agents contribute to power supply, data processing, or have direct interaction with the environment. Such testbeds allow researchers to address challenges in coordinating agents with differing dynamics and hardware limitations while optimizing system efficiency and robustness.

A UAV–UGV cooperative heterogeneous system was developed in [12], where the TMUAV is tethered to a UGV that houses the drone’s battery and includes a landing platform for compact transportation before deployment. This design allows the TMUAV to operate with an extended range and reduced weight, as the UGV supplies power and provides a stable base for takeoff and landing. The UGV not only extends the operational time of the UAV by serving as a mobile power station but also offers a stable platform for precise UAV deployments in scenarios such as search and rescue missions, infrastructure inspection, or remote data collection.

Another cooperative UGV-UAV system proposed in [63] focuses on mapping obstacle-filled environments and finding optimal paths. In this experiment, a simple 3 m fishing line was used as the tether instead of traditional communication or power cables. This lightweight tethering approach facilitated quick deployment and reduced the overall system weight, allowing for greater maneuverability of the UAV. The system is designed so that the UAV only assists the UGV when obstacles are too large for the ground robot to make autonomous pathfinding decisions. For instance, when encountering tall barriers or dense vegetation, the UAV provides an aerial perspective to help the UGV find alternative routes or identify navigable paths.

5. Variable Length TMUAVs

As discussed in the previous sections, TMUAVs offer enhanced versatility over other methods. The tether system enables these UAVs to transport payloads unsuitable for traditional gripping methods and is particularly effective for deploying sensors onto surfaces that are otherwise inaccessible, such as damaged structures. Applications like payload transportation and sensor deployment can benefit greatly from a retractable tether system attached to an MUAV, as shown in Figure 3. For instance, an MUAV equipped with a retractable tether can safely deploy sensors onto a damaged structure [64] without needing to approach it closely, reducing the risk of collision or further damage. However, incorporating a retractable tether system introduces significant challenges in controlling the MUAV. The coupled dynamics between the UAV and the cable system complicate modeling, and the time-varying dynamics during tether deployment or retraction must also be addressed. Additionally, the elongated tether is susceptible to external disturbances, such as wind-induced swinging, which can destabilize the entire system. Therefore, a robust control architecture is essential to manage the time-varying dynamics, mitigate the effects of tether swinging, and ensure stability. Furthermore, the faster dynamics involved in altitude control must be accounted for in scenarios such as operating in environments with limited clearance.

Retractable tether systems that is subject to smaller forces and high disturbances can be found in satellite systems. In [65], a MATLAB-based simulation environment was developed to test a tethered satellite system, focusing on the dynamic behavior of tethers in various deployment scenarios. The study modeled three distinct systems: a constant-length tether in Earth’s gravity, a variable-length tether in Earth’s gravity, and a variable-length tether under orbital conditions. This work provided a robust modeling approach for variable-length tethers in low-gravity environments, offering valuable insights applicable to TMUAV systems where a tether of changing length is employed. Additionally, Ref. [66] examined the challenges of tethered satellite systems for applications such as space debris cleanup and performing orbital maneuvers that would otherwise be difficult or impossible. One of the significant challenges identified was managing transverse and longitudinal vibrations along the tether, particularly in systems utilizing relatively long tethers. These vibrations become more pronounced during deployment, potentially destabilizing the system. To address this issue, the authors developed a numerical model and performed an in-depth analysis to devise a stable deployment strategy. Their proposed solution included implementing a control torque and managing transverse boundary displacement to minimize vibrations and enhance stability during tether deployment. The insights gained from these studies not only advance the field of satellite technology but also provide a foundational understanding for developing retractable tether systems in TMUAV applications. By adapting similar control strategies and modeling techniques, TMUAV systems can benefit from enhanced stability and versatility, particularly in missions that involve variable-length tethers in dynamic or constrained environments.

For slung payload transportation, Ref. [67] proposed a system for in-flight control of a variable-length tether. This system allows the payload to be lowered and raised without requiring the quadrotor to land, enabling safe operation in cluttered environments or rough terrain. The researchers developed a new optimization-based trajectory planner that manages input and state constraints while incorporating collision avoidance strategies. The system was tested in both simulation and physical lab environments. The experimental results demonstrated that the variable tether system could transport loads in scenarios where conventional systems would struggle. The ability to adjust the tether length dynamically improved tracking performance by reducing the impact of cable swing on the UAV’s stability. Although rigidly attached payload systems still outperformed the variable tether system in most performance metrics, the retractable tether offers significant advantages in versatility. Specifically, it enhances the UAV’s capability to perform tasks such as delivering emergency supplies in challenging or hard-to-reach locations, where precise load placement is critical.

Ref. [68] proposed an image-based visual serving control scheme for aerial transportation systems with variable-length cables, using a generalized virtual image feature. The approach utilizes onboard monocular cameras and constructs a feature signal that integrates image moments, cable length, and payload swing angles. A hierarchical anti-swing controller is developed to achieve payload swing suppression, precise cable length adjustment, and accurate image-based position tracking. Hardware experiments validate the effectiveness of the proposed controller in both the swing suppression and tracking of moving targets.

Ref. [69] proposed real-time trajectory generation for a quadrotor UAV with a pulley system based on model predictive control. The computational efficiency and performance of the system for navigating through tunnels were numerically validated. An anti-swing control and trajectory planning method for a variable-length cable TMUAV was proposed in [70]. The approach involves a coupled dynamic model that integrates the quadrotor’s dynamics under payload pull and the pendulum dynamics of the payload with a variable-length cable influenced by the quadrotor’s acceleration. An integral backstepping control scheme is designed to minimize payload swing, enhancing stability and control during flight. The key literature for varying length TMUAV is summarized in

Table 3. Key literature for variable length tether multirotor UAV systems.

Literature	Assumptions on Modeling	Contributions
Zeng et al. (2019) [41]	· Diameter of the pulley is much smaller than the cable length · Slack and taut conditions considered	· Differential flatness of the quadrotor–pulley–load system · Geometric controller for the quadrotor–pulley–load system
Oh et al. (2022) [69]	· Rigid rod model · Tether remains taut	· Real-time trajectory generation for the quadrotor–pulley system · Obstacle avoidance considered
Yang et al. (2022) [70]	· Cable mass is neglected · Tether remains taut	· Newton–Euler-based model of a quadrotor with a variable-length suspended payload · Cascade control strategy with integrated trajectory planning
Li et al. (2024) [67]	· Winch diameter is neglected · Cable is massless, inextensible, and remains taut	· Optimization-based motion planning for constant and variable-length payload transport in cluttered environments · Comparative analysis of three quadrotor-based transportation systems
Yu et al. (2024) [68]	· Payload is always positioned under the quadrotor · Cable is massless	· Dynamic relationship established between image features, cable length, and payload swing angles · Generalized virtual image feature used for controller design to ensure system stability

.

A geometric controller could be effectively applied to manage the dynamics of a retractable TMUAV system. Geometric control leverages concepts from differential geometry to address complex system dynamics without relying on traditional coordinate-based representations. Instead, it utilizes the intrinsic properties of the system, allowing the UAV movement to be controlled directly in the rotational space. This approach eliminates the need for linearization or approximation through a fixed coordinate system, enabling more precise and aggressive maneuvers in three-dimensional space. The benefits of geometric control are particularly valuable in retractable TMUAV systems, where variable tether lengths introduce complex and non-linear tether dynamics. Geometric control offers robust trajectory tracking even under significant disturbances, which is critical when operating in confined spaces with minimal margin for error. The ability to handle the full nonlinear dynamics of both the UAV and the tether system ensures stability and responsiveness during tether deployment and retraction.

Previous studies have demonstrated the effectiveness of geometric control in TMUAV slung payload transportation scenarios. Research such as [21,22,32,37,38,40,50,71] showed that geometric controllers could maintain system stability, suppress payload oscillations, and improve tracking performance in dynamic and challenging environments. These works highlight the potential of geometric control to enhance the operational flexibility and safety of TMUAV systems, especially when combined with a variable-length tether mechanism. Furthermore, [41] proposed a geometric controller for pulley system attached to a quadrotor which considers the torque and the reactive moments coming from the pulley. The proposed framework was validated through numerical simulation but not experimentally tested yet.

6. Conclusions

In this paper, we presented a comprehensive review of TMUAV systems, exploring their applications, challenges, and future potential. We began by examining the diverse applications of TMUAVs across different fields, identifying three primary use cases: payload transportation, disaster management, and data collection. These applications demonstrate how TMUAVs offer versatile solutions to a broad spectrum of engineering challenges, particularly in dynamic and constrained environments. We then delved into the primary challenges associated with TMUAV systems, focusing on the intrinsic physical limitations and complex dynamics of tethered operations. Specifically, we highlighted the challenges posed by tether dynamics, including robustness issues, impulsive forces resulting from slack-to-taut transitions, and the risk of tether entanglement. Additional challenges related to multi-TMUAV systems were also discussed, such as the complexities of cooperative control and coordination in multi-agent scenarios. The paper also reviewed existing testbed environments used for both numerical simulations and experimental setups. We covered experimental testbeds designed for single-TMUAV systems, cooperative multi-TMUAV operations, and heterogeneous multi-agent systems that incorporate TMUAVs alongside ground or maritime agents. These testbeds play a crucial role in validating control strategies and advancing the development of practical TMUAV applications. Finally, we explored the future potential of variable-length TMUAV systems utilizing retractable tethers. The introduction of retractable tether technology could significantly enhance the versatility of TMUAV systems, allowing for more adaptive and flexible operations in challenging environments. This innovation opens new opportunities for applications requiring precise control and maneuverability, such as delivering emergency supplies in rough terrains, performing complex inspection tasks, and executing advanced multi-agent coordination missions.

In this study, we identified future research questions that involves developing a robust control architecture for a retractable TMUAV system. First, we need system analysis framework for the coupled dynamics of the MUAV with the retractable tether system such as pulley system. Different modelings of tether dynamics will be tested and its performance and computationally efficiency will be compared. Then, we aim to design an adaptive geometric controller for that ensures safe and reliable system operation under diverse and challenging conditions. This involves accounting for the complex dynamics introduced by variable tether lengths, such as changes in system inertia, dynamic stability challenges, and managing impulsive forces during tether retraction and extension.

Additionally, building a dedicated testbed to generate experimental datasets for MUAVs with varying tether lengths is an important step. This testbed will facilitate real-world testing and validation of control algorithms, helping bridge the gap between simulation-based research and practical deployment. Experimental data will be invaluable for refining control strategies and enhancing the adaptability of the system. The potential applications of a retractable TMUAV system include sensor deployment in hard-to-reach areas, precise payload transportation in dynamic environments, and safe navigation through constrained or hostile environments. By integrating a retractable tether, the system can achieve higher versatility, enabling new capabilities such as delivering supplies in disaster zones and navigating in confined unknown spaces. To construct a testbed for this system, we plan to investigate the regulatory landscape related to the operation of TMUAVs in public spaces.