⚡ electrical engineering

Manipulation of Elasto-Flexible Cables with Single or Multiple UAVs

This paper proposes and validates a discretized model for multiple quadrotors manipulating deformable, extensible cables, demonstrating the system's differential flatness and the effectiveness of a closed-loop controller through numerical simulations and experimental tests.

Original authors: Chiara Gabellieri, Lars Teeuwen, Yaolei Shen, Antonio Franchi

Published 2026-07-30

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Chiara Gabellieri, Lars Teeuwen, Yaolei Shen, Antonio Franchi

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a world where robots aren't just rigid metal boxes that move in straight lines, but agile flyers capable of handling the messy, wiggly things in our world. This is the realm of aerial robotics, a field where drones (or Uncrewed Aerial Vehicles, known as UAVs) are being taught to do more than just take photos or deliver pizza. They are being trained to grab, drag, and shape objects that bend, stretch, and flop around. Think of a drone trying to untangle a fishing net, pull a fire hose, or scoop up floating trash from a river. The problem is that these "deformable" objects are incredibly hard to control. Unlike a solid brick, a rope or a cable doesn't have a single, predictable shape; it changes form the moment you tug on it. If you try to fly a drone while dragging a heavy, floppy cable, the cable swings, stretches, and twists, often pulling the drone off course. Scientists have long struggled to write the math that tells a drone how to move so the cable follows exactly where it's supposed to go without turning into a chaotic mess.

This paper tackles that exact challenge: how to make one or many drones work together to manipulate a stretchy, bendy cable. The authors propose a clever way to break the cable down into a chain of invisible beads connected by tiny, stretchy springs. By treating the cable this way, they discovered a mathematical "cheat code" called differential flatness. In simple terms, this means that if you know where a few specific points on the cable are (and how fast they are moving), you can calculate exactly what the drones need to do to get there. It's like knowing the path of the tip of a whip allows you to figure out exactly how to flick your wrist to make the whole thing dance. The researchers didn't just dream this up; they built a computer model to prove the math works, and then they went into a lab with real drones and a real cable to see if it held up in the messy, unpredictable real world.

The "Beads on a String" Strategy

The core idea of this work is to stop trying to model a cable as one continuous, smooth piece of rubber. Instead, the authors imagine the cable as a necklace made of heavy beads (called "point masses") linked together by bouncy, stretchy springs. Some of these beads are attached directly to the drones, while others just hang in the air. This "discretized" model turns a complex, wiggly problem into a series of simple physics puzzles: gravity pulls the beads down, the springs stretch and pull them back, and the drones push and pull on the beads they are holding.

The team proved mathematically that for a huge variety of setups—whether a drone is holding one end of a cable while the other end is tied to the ground, or two drones are holding opposite ends of a cable floating in mid-air—you can always find a set of "flat outputs." These are just specific points on the cable (like the beads right next to the drones) whose positions and movements tell you everything you need to know about the whole system. If you tell the computer, "I want these specific beads to move in a circle," the math automatically figures out the exact thrust and angle every single drone needs to use to make that happen, even accounting for the cable stretching and swinging.

From Computer Dreams to Real-World Flights

To test if this "beads and springs" idea was just a pretty simulation or something that actually works, the researchers set up a series of experiments. First, they needed to teach their computer model what the real cable looked like. They took a 1-meter long cable weighing about 7 grams and manually wiggled it around while a high-speed camera tracked its every move. They fed this data into their model to "tune" the springs and friction settings, essentially teaching the computer how stretchy and heavy their specific cable was. The result was impressive: the computer's prediction of where the cable would be was off by less than 2 centimeters on average. That's a tiny margin of error for something that is constantly flopping around.

Once the model was tuned, they put it to the test with two real drones (Crazyflie 2.1s). They made the drones fly in different patterns: some were simple straight lines, while others were complex 3D figure-eights. In these tests, the drones tried to move the cable along a specific path. The results showed that the model was good at predicting the cable's behavior, even when the drones were flying fast or the cable was stretched tight. The errors were small, mostly due to the natural lag in the drones' own motors and the slight imperfections in the camera tracking, but the cable generally followed the plan.

Taming the Chaos with Feedback

The most exciting part of the paper came when they added a "closed-loop" controller. This is like giving the system a pair of eyes and a brain that reacts in real-time. In a normal flight, if the cable stretches more than expected, the drone might overshoot its target. But with this new controller, the system constantly checks where the cable actually is versus where it should be. If there's a mistake, the controller instantly recalculates the drone's path to correct it.

They tested this by swapping out the cable for a different one with slightly different weight and stretchiness—something the computer model didn't know about. Without the feedback loop, the drones struggled, missing their target by over 30 centimeters. But once they turned on the feedback controller, the error dropped dramatically to around 10 centimeters. It was as if the system suddenly realized, "Hey, this rope is heavier than I thought!" and adjusted its flight plan on the fly to compensate.

What This Means

This work doesn't claim to have solved every problem in aerial robotics, but it takes a massive step forward. It proves that you can use a relatively simple "beads and springs" model to control complex, stretchy cables with drones, and that this model works well enough to be used in real-time control. The authors showed that by focusing on the right points of the cable, you can predict and control the whole system's behavior. While the math is complex, the result is a step toward a future where drones could autonomously handle hoses for firefighting, collect floating debris from waterways, or even repair underwater cables, all while keeping a firm, flexible grip on the task at hand. The paper suggests that with better models and more sophisticated control, these wiggly, unpredictable objects might soon become just another tool in a drone's toolkit.

Technical Summary: Manipulation of Elasto-Flexible Cables with Single or Multiple UAVs

Problem Statement
This work addresses the challenge of aerial robotic manipulation involving deformable and extensible cables. While traditional aerial manipulation often treats cables as rigid or purely elastic tools for suspended payloads, real-world applications (such as fire fighting, fluid conduit handling, and waterway maintenance) require managing cables that exhibit both bending flexibility and length variation (extensibility). Existing literature has largely focused on rigid cables, inflexible extensible models, or flexible inextensible models. The specific domain of flexible and extensible cable models in aerial manipulation remains under-explored, particularly regarding the theoretical properties required for control and trajectory generation.

Methodology
The authors propose a discretized representation of the cable, modeling it as a series of $n$ point masses interconnected by linear springs through passive spherical joints. This approach captures both the bending and the elastic elongation of the cable. The system dynamics are derived for three general classes of configurations:

Class (a): One cable end attached to the ground, with multiple quadrotors attached along the cable.
Class (b): Free-flying systems with one hanging end and the other attached to the last quadrotor.
Class (c): Systems where both cable ends are attached to quadrotors (including the widely studied two-robot case).

The core theoretical contribution is the proof of differential flatness for these systems. The authors demonstrate that for all three classes, a specific set of flat outputs exists. These outputs consist of the positions of specific points on the cable (specifically, the hanging end and the points immediately following each quadrotor attachment) and the yaw angles of the quadrotors. By establishing flatness, the authors show that all system states and control inputs can be expressed as functions of these flat outputs and their derivatives. This property enables the generation of dynamically feasible trajectories for the entire system.

Key Contributions

Theoretical Extension: The work extends the demonstration of differential flatness beyond the two-UAV case analyzed in prior literature to a large class of tethered multi-UAV systems. It incorporates cable elasticity, a factor often ignored in previous models but critical for real-world scenarios.
Flat Output Identification: The paper identifies meaningful flat outputs (specific cable point positions and robot yaw angles) that allow for the analytical computation of the full system state and control inputs.
Experimental Validation: Filling a gap in existing research, this work provides the first experimental validation of a discretized elastic cable model in an aerial manipulation system. The model parameters (stiffness, rest lengths, viscous friction) were identified from real-world data.
Closed-Loop Control: A model-based manipulation strategy is experimentally tested. This controller closes the loop on the cable output, utilizing feedback to correct for parameter uncertainties and tracking errors.

Results

Parameter Identification: Using a motion capture system, the authors identified cable parameters (spring coefficients and viscous friction) from a dataset where the cable was manually perturbed. The identified model showed an average position error of less than 0.02m compared to measured data.
Simulation: Numerical simulations confirmed that flatness-based trajectory generation allows the system to follow desired paths, though small tracking errors were observed due to initial conditions and numerical differentiation requirements.
Experimental Validation:
- Model Accuracy: In validation tests involving two Crazyflie 2.1 UAVs manipulating a cable with various trajectories (unidirectional and 3D figure-eight) and speeds, the model accurately predicted cable dynamics. The highest average output error recorded was 0.071m (in a narrow, fast configuration), with most errors remaining below 0.02m.
- Control Performance: Closed-loop experiments demonstrated the controller's ability to significantly reduce tracking errors in the presence of parameter uncertainties (e.g., using a different cable or different attachment points). For instance, in one test, the average tracking error for a specific cable point dropped from 0.373m (open-loop) to 0.108m (closed-loop).

Significance and Claims
The paper claims to establish a foundational framework for controlling multi-robot systems manipulating complex, deformable objects. By proving differential flatness for flexible and extensible cables, the authors enable the use of trajectory generation techniques that were previously unavailable for such systems. The experimental results validate that the proposed discretized model is sufficiently accurate to describe real-world cable behavior and can serve as the basis for effective feedback control laws. The work suggests that this approach can be extended to more sophisticated model-based controls and stability proofs in future research, as well as to the optimization of cable segment discretization and the manipulation of other complex deformable objects.

The "Beads on a String" Strategy

From Computer Dreams to Real-World Flights

Taming the Chaos with Feedback

What This Means

Technical Summary: Manipulation of Elasto-Flexible Cables with Single or Multiple UAVs

More like this