Biologically inspired drone immobilisation
Finding the most effective way to stop a drone, through 109 experiments
/images/drone-immobilisation/hero.webp
Rendered concept visual: an interceptor drone approaching a target drone entangled in cord under a parachute
- Client
- Master's thesis (MSc), 30 ECTS · Technical University of Denmark
- Context
- Supervisors: Torben Anker Lenau, David Bue Pedersen
- Year
- 2026
- Role
- Sole author
- Grade
- 12
- Tools
- Build-Measure-Learn loops · Biologically inspired design · Affinity mapping · Semi-structured interviews · High-speed video (240 fps) · Laser tachometer · FDM printing · Stakeholder mapping · Double diamond · Autodesk Fusion
- Contribution
- Sole author, designed, built and ran all 109 experiments
Master's thesis. Illegal drone incidents in Denmark rose from 62 in 2023 to 206 in 2024, however Danish police still have no local counter-drone tool other than a firearm, and this means that they usually decline to act, since the available response is either too slow or disproportionate to the threat. I interviewed 11 people across the Danish police, military and emergency management and mapped 373 quotes into an affinity map in order to establish what a usable tool would actually have to do, and then ran 109 experiments against a spinning propeller in order to find out what physically stops a drone without destroying it. The answer turned out to be a frayed rope, with five out of five instant stops on the rig, confirmed against a drone in free flight.
Why this question
The project started after the Copenhagen airport incident in September 2025, and the reason for looking at a physical solution rather than an electronic one is that jammers are being outpaced by autonomous drones and are already defeated by fibre-optic control in armed conflicts. This leaves a physical gap, since there is currently no non-destructive way to bring down a 250 g–2 kg quadcopter regardless of how it is being flown. The thesis combines the double diamond with problem-driven biologically inspired design, which means the first half was spent finding out what the tool actually has to do before any of it was tested.
The empirical research
Eleven participants contributed, with nine long-form semi-structured interviews plus two shorter conversations with qualified people, covering the Danish police, the military and emergency management, and supplemented with drone hobbyists in order to get the perspective of the people being regulated rather than only the people regulating. The interview design drew on The Mom Test, which means asking about specific past incidents rather than about opinions on the future, since what people say they would do is considerably less reliable than what they have already done. Participants are anonymised and coded P1 to P11 throughout. The analysis was an affinity mapping of 373 quotes, taken from the transcripts onto post-it notes and then clustered into 16 groups, after which the clusters were named and connected with drawn lines of cause and effect. This is the part of the work that decided everything downstream, since the requirements the experiments were later run against came out of these clusters rather than out of my own assumptions about what a counter-drone tool should be.
What the research actually found
Detection came out as the central cluster, and this was a surprise, since the thesis started from the assumption that the gap was in the response rather than in knowing there was something to respond to. The police described having no overview and wasting resources chasing what one participant called ghost reports, which became especially common after the airport incident, and one noted that many false reports turned out to be civilian aircraft identified afterwards using flight radar. Here the underlying issue is identification rather than detection, since a drone has no licence plate, which means an officer cannot tell whether they are looking at a hostile operator or at a municipality photographing a property. The second cluster was collateral damage, and it explains the behaviour the whole project exists to change. Shooting a drone down produces debris that lands on buildings and potentially on people, and one participant put it plainly that even a successful shoot-down still results in something falling, which is why police frequently decide that taking the drone down would cause more harm than leaving it alone. Furthermore one participant made the point that getting a drone down intact carries forensic value for the investigation afterwards, and that he could see far more value in having the drone in their hands than in it smashing into something and going into a thousand pieces. The third cluster was governance, where the recurring theme was fractured responsibility. Denmark has 12 police districts, each using drones in its own way, with no common direction, no joint procurement and no shared approach to training, while responsibility splits at altitude, since the police hold it up to 200 metres and the military above that. It is important to note that this was described as being actively worked on, with a national situational awareness initiative started after the airport incident.
Turning the research into requirements
The affinity map produced a stakeholder map showing that regulatory, operational and legislative authority each sit in different places, and that the police carry the majority of the operational responsibility during peacetime while the picture would shift toward the military in wartime. This is what established who the tool is actually designed for. Furthermore the research produced a framing of the problem I would not have arrived at alone, being that illegal drone operators divide into the clueless, the careless, the curious and the criminal, which matters because a response proportionate to a criminal is wildly disproportionate to someone who simply did not know the rules. Here this is the argument for a non-destructive tool rather than a faster destructive one, since a police officer facing an unidentified drone cannot know which of the four they are dealing with, and a tool that brings the drone down intact removes the need to decide before acting. The requirement that came out of the first diamond was therefore narrower than the one the project started with. Not a way to stop a drone, and instead a way to stop a 250 g to 2 kg quadcopter without destroying it, without producing falling debris, and regardless of how it is being controlled, since fibre-optic and autonomous drones defeat everything electronic.
How I tested
The rig was a brushless drone motor with a 7-inch propeller and an ESC on PWM control, with RPM verified by laser tachometer at roughly 10,000, which is equivalent to a hovering camera drone. Test objects were dropped into the spinning propeller and measured on RPM change, clean stops and entanglement, alongside slow-motion video at 240 fps. It is important to note that every test object began as a written, falsifiable hypothesis, and that the three Build-Measure-Learn loops covering 109 experiments were deliberately run for breadth rather than statistical significance, since only the promising results were repeated. Furthermore all of the CAD for this project was done in Autodesk Fusion, which came to more than 40 models by the end.
The error I had to fix mid-way
The acrylic safety enclosure turned out to be creating turbulence and causing test objects to ricochet off the walls back into the propeller, which was inflating the results. This is due to the enclosure restricting the airflow and giving ejected objects nowhere to go, and remounting the rig on a wooden pallet solved both problems at once. Furthermore every earlier result had to be re-read against that, since the ones recorded before the change could not be trusted in the same way.
What the experiments found
Phase 1, at 18 experiments, killed my starting assumption, since prior work suggested that looped strings were effective and that more loops would be better, however the loops correlated negatively with entanglement due to thrusting each other away on contact. The one exception was a spiderweb-inspired hexagon, which entangled and stopped the motor cleanly. Phase 2, at 38 experiments, then rejected its own hypothesis, as rigid hexagonal frames reached 75% entanglement at 6 of 8 but produced no full stops, only RPM reduction correlating with frame diameter, while magnets and fishing hooks did nothing at all. Here the real finding came by accident, since a magnet fell below the rig with twine attached above it, and the weight stopped the propeller flinging the string clear, which locked the motor. This became the drop-line principle. Furthermore the phase produced the contrast that mattered most, as 0.11 mm braided line entangled 60% of the time but was always cut, while 2 mm twine entangled only 9% of the time and stopped the motor nearly every time it did.
The resolution
Phase 3 reframed the problem after the Cape sundew, which catches insects with a fast adhesive and then closes slowly and mechanically, and instead of hunting for one optimal string I tested a two-part entangler-plus-stopper. This worked, however the standout result was simpler, since a 6 mm rope unravelled at the end stopped the motor in 5 of 5 trials, instantly, with no ricochet and even at full throttle, while a single strand from the same rope did nothing. This is due to the many fine frayed strands acting as the entangler and presenting too many contact points for the propeller to clear, which drags the solid rope into the motor where friction stops it.
Validating on a real drone
Static-rig results mean little if they do not transfer, so I removed the enclosure, tested from a 2.5 m staircase, and then tested against a custom FPV quadcopter in free flight, capable of 100 km/h at roughly 12,000 RPM. The frayed rope and the braided-line-plus-twine combination both stopped the propulsion instantly, and the frayed rope entangled two propellers at once. However the most useful finding was not about the material at all, since hitting the drone by hand was hard, and this means that area of effect matters more than elegance, which is what drove the concept toward an interceptor drone delivery system with a parachute release, so that the target descends under control instead of falling.
Design requirements derived
The entangler needs high shear strength, a small diameter below 0.3 mm and high flexibility, while the stopper needs high shear strength, high tensile strength, a larger diameter above 2 mm and a high-friction surface. Furthermore the delivery method needs a significant area of effect, since that turned out to be the limiting factor rather than the material itself.
Limits
It is important to note that the static rig was bolted down, and this means that the destabilisation principles could not be verified until the free-flight tests. Furthermore the sample sizes were small by design, since the loops were run for breadth, and the concept seems to have a real weakness against agile drones with collision avoidance and motor guards.
/images/drone-immobilisation/affinity.webp
Digital affinity map clustering interview findings into themes: counter-drone measures, detection, threats, regulation
/images/drone-immobilisation/test-rig.webp
Brushless motor and propeller test rig inside a clear acrylic enclosure
/images/drone-immobilisation/test-objects.webp
Three-panel bench test: frayed line approaching, entangling and releasing from a mounted propeller
/images/drone-immobilisation/drone-validation.webp
Top-down photo of the FPV quadcopter with a frayed blue line entangling two propellers
/images/drone-immobilisation/concept-diagram.webp
Four-panel diagram of the interceptor aiming, releasing twine, entangling the target and detaching by parachute