We are building Europe's drone wall.

A mesh of optical M0THER carriers, releasing hardened single-use S0N micro-interceptors against FPV swarms and Shahed-class loitering munitions. Sized from a nine-soldier perimeter to a national frontier.

Cluj-Napoca · Romania Pre-prototype · 2026
§01

Doctrine

Five claims · 7 min read

An argument for how counter-drone has to be fought, in five claims.

The first time a four-hundred-euro quadrotor disabled a forty-million-euro main battle tank, the cost geometry of land combat changed and did not change back. The same decade, fifty-thousand-euro Iranian-pattern loitering munitions began crossing borders by the hundred to strike substations and city blocks a thousand kilometres from where they launched — including, repeatedly, into NATO airspace. The two ends of that spectrum — the squad-scale FPV and the frontier-scale Shahed — share a single property: they are cheap enough to be expended in numbers no defender can match per-unit. Counter-drone is the layer of warfare that did not exist a decade ago, and that no European force currently fields at the tempo the threat actually arrives at, at either scale.1

The obvious response is to apply legacy air-defence thinking to the new problem, and this has been the response most Western primes have offered. It is the wrong response. Air defence was designed against expensive, scarce, attributable threats; counter-drone has to win against cheap, abundant, anonymous threats — at squad scale against FPVs, and at frontier scale against loitering munitions, with one doctrine answering both. The economics do not match. Neither does the doctrine that produced the economics. What follows are five claims about how counter-drone has to be fought instead, in the order an honest engineer working on the problem tends to discover them.

§ I

Passive over active.

Active detection — radar — announces the position of the radar. Against an adversary that fields a three-hundred-euro drone with a kamikaze warhead, this trade is unaffordable. The M0THER detects by watching, on station, at altitude. Classification runs on the M0THER itself. The detection layer emits nothing — there is nothing to locate, nothing to home in on.

Optical detection trades omnidirectional reach for survivability. A camera receives photons; a radar emits them. The first cannot be located by what it does not radiate; the second can. We accept the trade because three M0THERs in overlapping patrol orbits paint a continuous field, and they do it while emitting nothing.

The choice of optical-only on a noisy carrier platform is a deliberate one. A microphone array on a M0THER would hear its own rotors first and the threat second. Visible-light and thermal apertures hear nothing — and from altitude, the geometry is on their side.

§ II

Distributed over centralised.

A network has a centre or it does not. Centralised command and control was a defensible design choice in an era when the cost of compute and communications dictated star topologies, and when the adversary lacked precision-guided answers to the centre. Both conditions have ended. A modern adversary will find the command post and strike it inside the same hour the system is detected.

The fleet has no command post. Three M0THERs over a position coordinate over a peer mesh and reach consensus on a target without privileged authority. The operator's tablet, when present, is an observer and a lock-cueing device, not a control point. Removing the tablet does not disable the system. Removing one M0THER degrades coverage but does not break the consensus floor. The fleet behaves as a network of agreeing peers, and the property a network of agreeing peers has, that a star topology does not, is that there is nothing in particular to destroy.

Legacy command and control C2 single point of failure Distributed peer mesh no root · degrades, does not break
Fig. 1Legacy centralised command versus the distributed peer mesh. The root in the left-hand topology is the target an opponent will pay most to find; the mesh on the right has no root.
§ III

Single-shot over reusable.

The S0N is a single-use drone. Each S0N launches once, intercepts once, and does not return. YOLO — you only launch once — is the company's name because it is the company's first design constraint. Everything else in this claim follows from accepting that the interceptor is expended. Reusable interceptors are mechanically expensive, electrically complex, and require maintenance cycles that a forward squad does not own. A hardened micro-interceptor — palm-sized, fast enough to close on a target inside seconds, perched aboard a M0THER or forward-cached on the ground — is built to survive one impact and no maintenance cycles. The economic floor of the system is the S0N, not the M0THER.

This inverts the cost geometry that has defined air defence for forty years. A Patriot battery firing a four-million-euro effector at a thirty-thousand-euro Shahed is winning the engagement and losing the war. A S0N costing a fraction of its target, dispatched by a M0THER that pays for itself across many engagements, is fighting a fight the defender can sustain.2

§ IV

Edge over cloud.

Every component of the detect–classify–dispatch chain runs on the M0THER itself. Detection runs on-device. Classification runs on the same compute. Mesh consensus runs across local radios that need no cellular tower, no satellite, and no off-platform compute. Once cued, the S0N runs its own terminal guidance to the target. The system has no backhaul because it requires none.

Cloud architectures presume connectivity. Counter-drone operates in environments where connectivity is the first thing an adversary removes. A doctrine that assumes the cloud will be there is a doctrine that ends at the first jammer. Edge-only is not a feature of the design — it is a precondition for the doctrine to function.

§ V

Modular over universal.

Universal weapons are a procurement convenience, not an operational virtue. Threats vary, and the responses must vary with them. Each M0THER carries a mixed S0N loadout — kinetic and fragmentation effectors, day-optimised and night-optimised — drawn from a forward-cached canister and reloaded in field under thirty seconds. The mechanical interface is keyed and instantaneous. A unit rebalances its swarm composition by mission, not by procurement cycle.

Doctrine, in the end, is what survives encounter with cost. Each of the five claims above is also a budget decision. Together they describe an economics the defender can afford and an opponent cannot exhaust.

The system described next is what the doctrine looks like once it is built. Treat it as the design, not the proof.3


  1. 1.European Defence Agency capability-gap reporting through 2026 places counter-UAS among the most under-fielded ground-force capabilities across EU member states. NATO airspace has logged repeated Shahed-debris recovery events along Romania's eastern border since 2022.
  2. 2.Cost figures shown are illustrative and component-level. Full system economics are detailed in the Architecture Brief, available under NDA.
  3. 3.All technical figures on this page reflect concept-architecture state. YOLO Machines is a pre-prototype company; figures are design targets, subject to engineering revision through the bench-prototype phase.
§02

System

From squad to frontier

Three M0THERs aloft. Twelve S0Ns ready.

A squad-scale deployment is three M0THER carriers in triangular patrol over a nine-soldier position, each carrying four S0N interceptors. The carriers watch — visible-light and thermal — across overlapping fields and require two-of-three agreement before any S0N is launched. From detection to airborne S0N: under two seconds. Terminal closure: seconds more. Carriers reload from forward-cached canisters between engagements. This is the architecture; the company is at the bench-prototype phase of building it.

consensus k=2 TGT-01 · inbound SQUAD · 9 M-01 M-02 RELEASING M-03 N 0 100 m Soldier M0THER S0N in flight
Fig. 2M0THER patrol over a squad position. Three carriers share an airspace picture across a peer mesh; M-01 and M-02 have independently classified the inbound target; consensus is satisfied; M-02 releases a flight of S0Ns.
Detect to airborneunder 2 s
M0THER endurance4 h on station
Sensor reach1.5 km · visible / thermal
M0THER cruise90 km/h · transit
S0N burst speed180 km/h
S0N effective range1.5 km · single charge
Consensus ruletwo of three M0THERs
Communicationspeer mesh · no backhaul

Autonomy

How the M0THER carrier thinks.

Detection, classification, mesh consensus, and dispatch authorisation are designed to execute on the M0THER itself. There is no cloud component, no central server, and no off-platform compute. This is the doctrine of § IV — edge over cloud — implemented in software.

Each carrier runs the same software. It watches, classifies what it sees against an onboard threat library, and shares those classifications across the peer mesh. A launch is authorised only when two of three carriers independently agree — a rule that lives on the mesh itself. No node holds authority over the others, and there is nothing off-platform for an adversary to cut.

PEERS · 6 NO ROOT CONSENSUS · k = 2 EDGE-ONLY
Fig. 3Topology of the autonomy mesh. No node is privileged; any can be lost without breaking the consensus floor.
  • OpticalWide-field visible and long-wave thermal apertures per M0THER. Threat library trained on field captures of contemporary FPV and loitering-munition signatures.
  • ConsensusFault-tolerant track formation across the M0THER peer mesh. Two-of-three agreement required before launch authorisation propagates.
  • Terminal guidanceOnboard imaging seeker on each S0N. Target lock from cue; sub-meter terminal accuracy on closing geometry.
  • Edge computeOn-device inference. No off-device computation. No backhaul. No cloud.
  • CommunicationsPeer mesh between M0THERs. No cellular, no satellite, no central server.

Modular S0N

Four S0N configurations. Mechanical interface. Field-swappable.

KIN · DAY KIN · NIGHT FRAG · DAY FRAG · NIGHT
Fig. 4The four S0N configurations. The mechanical interface is keyed and instantaneous; the choice is made in field, not at procurement.

Frontier mode

The same mesh, at national scale.

The squad configuration is the close-defence case. The same protocol — passive optical detection, peer consensus, edge compute, single-use intercept — extends to frontier scale by stretching the M0THER mesh into a continuous patrol along the line. Each M0THER station carries its own S0N flight, classifies inbound threats from altitude, and contributes to the same two-of-three consensus before any S0N is launched. The geometry changes; the doctrine does not. The drone wall conversation Eastern Europe has been having since 2024 is what this looks like when it is built.

FRONTIER consensus k=2 TGT · loitering munition M-01 M-02 M-03 M-04
Fig. 5Frontier-mode deployment. M0THERs in continuous patrol along the line, on a single mesh. Two carriers must agree before any S0N is launched.
§03

Programme

Build trajectory · 2025 — 2028

Architecture-locked. Bench prototype underway.

The system is at the architecture-locked stage. The next two years route through bench prototype, tethered flight, free flight, and live intercept demonstration, in that order. Series production follows. Phase 02 is funded; subsequent phases are open to capital and programme partners.

PHASE 01 PHASE 02 PHASE 03 PHASE 04 PHASE 05 PHASE 06 Architecture Bench prototype Tethered flight Free flight Live intercept Production 2025 — 26 Q1 2026 Q2 — Q3 2026 Q4 2027 Q2 2027 Q4 2028 NOW
Fig. 6Programme schedule. Six phases from architecture-locked through series production. Funded through Phase 02; subsequent phases dependent on partnership and capital.

Phase 01

Closed

Architecture

2025 — 2026 Q1

Concept-architecture, doctrine, modelling. Design-of-experiments for sensor fusion and consensus rules. Architecture Brief V1.

Phase 02

Active

Bench prototype

2026 Q2 — Q3

M0THER and S0N frames at the bench. Comms stack, autonomy stack, first end-to-end mesh consensus on simulated targets.

Phase 03

Planned

Tethered flight

2026 Q4

First M0THER hover and S0N release under tether. Mesh consensus exercised on live optical input.

Phase 04

Planned

Free flight

2027 Q2

Untethered M0THER patrol orbit. S0N free-flight envelope characterisation. Full three-node mesh with degraded-mode testing.

Phase 05

Planned

Live intercept

2027 Q4

Mesh consensus dispatch with terminal effect on instrumented target drones. Both squad and frontier configurations.

Phase 06

Planned

Series production

2028

Series manufacture and first fielding. Production line in Cluj-Napoca; supply chain inside the European Union.

§04

Contact

Currently pre-prototype

Open a conversation.

hello@yolomachines.com

General & press

Mark press in the subject line; same-day routing.

Investors & partners

Architecture Brief and engagement modelling available under NDA.