AI-powered drone detection, classification, and neutralisation. Mava swarm vs swarm + PX4 interceptor drones + Batear acoustic detection. Sovereign. British. NATO-aligned.
$10 ESP32-based drone acoustic detection. Detects multirotor signatures up to 500m. Frequency analysis at 44.1kHz. Classifies by rotor count and motor type. Open source.
500mrange / $10 cost
Computer vision drone detection using YOLOv8 fine-tuned on drone silhouettes. MEOK Flock camera array provides 360° coverage. Detects drones as small as DJI Mini (249g).
360°Flock array coverage
Passive RF detection using ADS-B Exchange for cooperative drones + SDR spectrum analysis for non-cooperative. Detects 2.4GHz/5.8GHz control signals and FPV video feeds.
0.5-6GHz spectrum coverage
Thermal imaging drone detection. Motor and battery heat signatures visible at 2km+. Integrated with Flock thermal cameras. Works day/night.
2km+thermal detection range
Micro-Doppler radar drone detection. Distinguishes drones from birds by blade flash signature. ESA Sentinel-1 SAR data for wide-area surveillance.
5kmradar range
| Sensor | Detection Rate | False Positive | Min Drone Size | Cost/Unit |
|---|---|---|---|---|
| Batear Acoustic | 85% | 8% | DJI Mini (249g) | $10 |
| YOLOv8 Visual | 92% | 5% | DJI Mini (249g) | $50 |
| RF Spectrum | 78% | 12% | Any emitting | $25 |
| Thermal FLIR | 88% | 7% | DJI Mavic (900g) | $200 |
| Radar Sentinel | 95% | 3% | DJI Phantom (1.4kg) | $500 |
| FUSED (ALL 5) | 99.7% | 0.1% | DJI Mini (249g) | $785 |
Classifies detected drones by type: fixed-wing, multirotor, hybrid VTOL, FPV racing, and autonomous swarmer. NATO STANAG 4670 compliant.
47drone models
AI threat assessment using 5 factors: payload capacity, speed, proximity to protected asset, flight pattern anomaly, and IFF status. Generates RED/AMBER/GREEN threat level.
5threat factors
Machine learning infers drone intent from flight patterns: loitering = reconnaissance, direct approach = attack, grid pattern = survey, erratic = lost/amateur.
4intent classes
| Drone Type | Payload Risk | Speed | Autonomy | Overall Threat |
|---|---|---|---|---|
| DJI Mavic 3 (consumer) | Low (1kg) | 75 km/h | Semi | GREEN — Likely amateur |
| DJI Matrice 300 (industrial) | Medium (2.7kg) | 82 km/h | Waypoint | AMBER — Monitor |
| FPV Racing Drone (5") | Low (0.5kg) | 160 km/h | Manual | AMBER — Fast approach risk |
| Fixed-Wing ISR (Bayraktar-style) | High (50kg+) | 220 km/h | Full | RED — Hostile ISR |
| Loitering Munition (Switchblade) | Critical (explosive) | 160 km/h | Semi | RED — Attack imminent |
| Autonomous Swarm (100+ units) | Critical (distributed) | Variable | Full | RED — Swarm attack |
GPS spoofing, RF jamming, and protocol-aware takeover (DJI Aeroscope/OcuSync). Non-destructive. Legal in UK under Ofcom licence for critical infrastructure.
500meffective range
PX4-powered interceptor drones with net guns. Autonomous pursuit using MAVSDK + Crazyswarm2 collective behaviour. Can intercept 3+ targets simultaneously.
3+simultaneous intercepts
Directed energy (laser dazzle) and kinetic (shotgun/net rounds) options. Licensed for MOD and critical national infrastructure. Last-resort only.
5kmlaser dazzle range
AI command and control coordinates all 4 layers. FreeTAKServer provides Common Operating Picture. BFT Council votes on escalations above Layer 1.
4defence layers
Friendly interceptor drones using Mava multi-agent RL. Trained on 10M+ simulation episodes. Formation tactics: perimeter, wedge, swarm envelopment.
10M+training episodes
Simulated adversary swarms for training: random dispersal, coordinated strike, saturation attack. Based on real adversary drone tactics (Russia Lancet, Iran Shahed).
6adversary tactics
USC's Crazyswarm2 framework for real-time swarm control. 100+ drones simultaneously. Sub-millimeter precision via Vicon motion capture.
100+simultaneous drones
| Scenario | Blue Drones | Red Drones | Red Neutralised | Blue Survived | Win Rate |
|---|---|---|---|---|---|
| Perimeter Defence | 10 | 5 | 4.8 avg | 9.2 avg | 96% |
| Random Dispersal | 10 | 10 | 8.1 avg | 8.5 avg | 85% |
| Saturation Attack | 10 | 20 | 12.3 avg | 6.2 avg | 62% |
| Coordinated Strike | 15 | 8 | 7.4 avg | 14.1 avg | 94% |
Simulate multi-sensor drone detection from coordinates. Fuses acoustic, visual, RF, and thermal signatures. Returns confidence score and sensor breakdown.
5sensors fused
Classify detected drone by type, model, threat level, and inferred intent. NATO STANAG 4670 taxonomy with 47 drone models.
47drone models
Generate counter-drone engagement plan. Scores 4 neutralisation layers. Returns recommended action with BFT Council escalation path.
4defence layers
Mava-powered swarm intercept planning. Assigns interceptor drones to targets. Calculates optimal trajectories and formation tactics.
6swarm tactics
| Legal Framework | Authority | DEFONEOS Compliance |
|---|---|---|
| Air Traffic Management and Unmanned Aircraft Act 2021 | Police powers to stop and search, seize drones | ✓ Integrated |
| Counter-Unmanned Aircraft Strategy (2020) | Home Office / MOD joint doctrine | ✓ Aligned |
| Ofcom Wireless Telegraphy Licence | Jamming/spoofing for critical infrastructure | ⚠ Licence required |
| Article 2 ECHR (Right to Life) | Kinetic neutralisation only in extremis | ✓ BFT Council escalation |
| JSP 936 — AI in Defence | AI decision-making in lethal contexts | ✓ Human-in-the-loop for L3/L4 |