libdrone Wing — Platform Draft
About¶
Concept and pre-design document for libdrone Wing — the fixed-wing survey companion platform. Designed for wildlife survey missions at dawn and dusk using thermal imaging. Shares the GX12-7 payload standard with libdrone Pro. Primary initial application: deer and wild boar population survey for Czech hunting associations.
libdrone Wing — Platform Draft V0.1¶
Date: March 2026 | Status: Concept / Pre-design Author: Jakub, Telč, Czech Republic Relationship: Companion platform to libdrone Pro 6" — same payload standard, different mission profile
THE PROBLEM¶
Česká obec myslivecká (Czech Hunters Association) — ~70,000 members — is under severe pressure from the state, agricultural oligarchs, and landowners to reduce deer and wild pig populations tenfold. They are being asked to deliver results they cannot currently quantify or defend with data.
The specific operational need: - Survey large areas of mixed terrain (fields, forest edges, orchards) at dawn or dusk - When thermal contrast between animals and ground is maximum - Count individuals, record GPS positions, produce a report - Repeat seasonally to demonstrate trend data over time
Why current tools fail them: - Hired survey helicopters: €3,000+/hour, impractical for routine use - Commercial survey drones (DJI): closed ecosystem, Chinese origin, NDAA/EU compliance issues, expensive - Manual ground counts: slow, incomplete, disturbs animals, no GPS data - Fixed ground cameras: point coverage only, not area survey
What they actually need: A service. Not hardware. A report on a desk that says: "Field X, dawn survey 6 April 2026, 23 deer confirmed at GPS coordinates attached, 4 wild pigs at coordinates attached." Repeatable, consistent, defensible in a ministry meeting.
THE SOLUTION: libdrone Wing¶
A commercial fixed-wing or flying wing airframe with: - H7A3-WING flight controller running ArduPilot - Thermal imaging payload in a proper belly bay - Autonomous waypoint grid survey mission - GPS-tagged thermal detection logging - GX12-7 payload interface (same standard as libdrone Pro) - HDZero video link for real-time monitoring (optional — see variants)
The pilot sits at a safe distance with a laptop running Mission Planner. The ranger watches the live feed and/or reviews the GPS-tagged detection log after landing. No heroic low-altitude flying. No dark treetop navigation. Autonomous grid at 40-60m AGL.
Key insight: this is not libdrone Pro flying a survey. This is a dedicated survey platform that shares the payload interface standard. The glider is the fast path to first revenue. libdrone Pro stays on its own development timeline without pressure.
MISSION PROFILE¶
Target survey area: 10-50 hectares per mission (typical Czech field/forest edge) Survey altitude: 40-60m AGL (above ground level) Time of operation: Dawn (30 min before to 60 min after sunrise) or dusk Thermal window: October–February (ground cool, animals warm, maximum contrast) Flight time required: 30-60 minutes per mission Swath width at 50m AGL with FLIR Boson 320: ~60-70m Passes for 20ha field at 60m swath: ~12 passes = approximately 20-25 minutes flight Operator count: 1 pilot (autonomous mission monitor) + 1 ranger (payload operator / spotter)
AIRFRAME OPTIONS¶
Option A — Skywalker X8 (recommended primary)¶
- Type: Flying wing, 2122mm span
- Payload bay: Large belly bay, ~800g payload capacity
- Endurance: 45-75 minutes depending on battery
- Availability: EU distributors, €150-200 kit
- ArduPilot support: Excellent, extensively documented community builds
- Notes: Most common survey platform in EU amateur/research community. Large prop gives good endurance. Belly bay already designed for payloads. Some assembly required.
- Weight with electronics: ~2.2-2.8kg AUW
- EASA category: Specific category likely (>900g) — operational authorisation needed
Option B — Finwing Penguin (dedicated survey)¶
- Type: Conventional fixed-wing, ~1400mm span
- Payload bay: Dedicated survey belly bay, clean integration
- Endurance: 45-60 minutes
- Availability: EU importers, ~€200-280
- ArduPilot support: Good, community documented
- Notes: Designed specifically for survey work. Better payload integration than X8. Less community support than X8 but purpose-built. Easier weight distribution management for thermal payload.
- EASA category: Likely A3 or Specific depending on final AUW
Option C — Mini Talon V2 (smaller, faster to fly)¶
- Type: Flying wing, 1300mm span
- Payload bay: Moderate, ~400g payload
- Endurance: 35-50 minutes
- Availability: EU distributors, ~€100-130
- ArduPilot support: Very good
- Notes: Smaller, cheaper, easier to transport in the van. Payload capacity may limit thermal camera choice to Lepton (cheaper, lower res) rather than Boson. Good for proof of concept phase.
- EASA category: A2/A3 depending on AUW — easier regulatory path
Recommendation¶
Phase 1 POC: Mini Talon V2 or Skywalker X8 with FLIR Lepton thermal Phase 2 operational: Finwing Penguin or Skywalker X8 with FLIR Boson 320 Start with Option C or A for the myslivci pilot. Prove the concept cheaply. Upgrade airframe if warranted by demand.
ELECTRONICS¶
Flight Controller¶
Matek H7A3-WING (already owned) - ArduPilot native - Supports fixed-wing, flying wing, VTOL - Waypoint navigation, RTH, autonomous grid missions - Telemetry via UART for Mission Planner ground station - ~€85 (already purchased)
GPS¶
Matek M10Q-5883 (shared SKU with libdrone Pro) - M10 GNSS, compass - ~€25
RC Link¶
ExpressLRS — RadioMaster TX16S transmitter (already owned) + receiver - Recommended receiver: RadioMaster RP3 or BetaFPV SuperD for fixed-wing range - ELRS protocol: open, long range, reliable - For survey missions at 500m-2km range: ELRS 868MHz preferred over 2.4GHz - ~€20-35 receiver
Video Link (two options)¶
Option V1 — HDZero (maintain ecosystem consistency) - HDZero Freestyle V2 or Whoop Lite VTX - HDZero Goggle 2 already owned — analog receiver built in - Range: adequate for survey distances if antenna positioned well - Pro: consistent with libdrone Pro ecosystem, same goggles - Con: not designed for long-range fixed-wing survey, may need antenna upgrade - ~€60-80 VTX
Option V2 — No live FPV video, telemetry only - Mission Planner on laptop shows aircraft position on map in real time - Thermal data logged to SD card, reviewed after landing - Simpler, lighter, longer range operation - Ranger reviews detections post-flight rather than live - Pro: cleaner, lighter, no video link complexity - Con: no live situational awareness for ranger
Recommendation: Start with V2 (telemetry only) for the POC. Add HDZero later if live feed proves operationally necessary. The GPS-tagged detection log is more useful to the myslivci than live video anyway.
Telemetry¶
SiK radio 433MHz or 915MHz — Mission Planner ground link - ~€25 for a pair (air unit + ground unit) - Ground unit plugs into laptop USB - Shows aircraft position, altitude, battery, flight mode in real time
ESC + Motor¶
Airframe-dependent — typically included with kit or specified by community builds. Example for Skywalker X8: - Motor: SunnySky X2216 or similar 1250KV - ESC: 40A BLHeli_S - ~€30-50 motor+ESC
Battery¶
6S LiPo 4000-6000mAh for X8/Penguin 4S LiPo 3000-4000mAh for Mini Talon Recommend: same 6S philosophy as libdrone Pro for charger/system consistency where possible. ~€40-70 depending on capacity
THERMAL PAYLOAD¶
Phase 1 POC — FLIR Lepton 3.5¶
- 160×120 thermal resolution
- 57° FOV
- ~9g module weight
- Connects via SPI to ESP32-S3
- GX12-7 interface (same as libdrone Pro Payload 2)
- ~€200-250 module
- Limitation: Low resolution means individual animal identification at altitude is marginal. Adequate for presence/absence and count at <50m AGL.
Phase 2 Operational — FLIR Boson 320¶
- 320×256 thermal resolution
- Multiple FOV options (50°, 32°, 18°)
- ~60-80g module weight
- Digital interface (MIPI or USB)
- ~€800-1,200 depending on lens
- Advantage: Clear individual identification at 50-60m AGL. 60m swath with 50° lens. Suitable for professional survey reports.
Detection Logging¶
ESP32-S3 companion board: - Reads thermal frame - Logs GPS-tagged JPEG thermal snapshot when trigger pressed (ranger button on Radiomaster) - Or: automatic detection via simple threshold (hot blob = animal candidate) — saves GPS + snapshot - Writes to MicroSD - GX12-7 payload interface connection - ~€8-12 module
Output Format¶
Post-flight processing script (Python, FOSS): - Input: GPS log + thermal snapshots - Output: KML/GeoJSON overlay (opens in Google Maps, QGIS, or any GIS) - Output: PDF report with map, detection count, GPS coordinates table - This is what goes on the ministry official's desk
SHOPPING LIST WITH ESTIMATED PRICES¶
| Item | Option | Est. Price (EUR) | Notes |
|---|---|---|---|
| Airframe | Mini Talon V2 | €100-130 | POC option |
| Airframe | Skywalker X8 | €150-200 | Better payload bay |
| Airframe | Finwing Penguin | €200-280 | Most professional |
| FC | Matek H7A3-WING | €0 | Already owned |
| GPS | Matek M10Q-5883 | €25 | Shared SKU with Pro |
| RC receiver | ELRS 868MHz | €25-35 | Long range |
| Telemetry | SiK 433MHz pair | €25 | Mission Planner link |
| Video | HDZero Whoop Lite | €60-80 | Optional V1 |
| Motor + ESC | Airframe dependent | €35-55 | Often kit-included |
| Battery 6S 5000mAh | LiPo | €45-70 | |
| Thermal: Phase 1 | FLIR Lepton 3.5 | €200-250 | POC |
| Thermal: Phase 2 | FLIR Boson 320 | €800-1200 | Operational |
| ESP32-S3 payload board | €8-12 | Detection logging | |
| GX12-7 connectors | Set | €15-20 | Shared with Pro |
| Miscellaneous | Cables, mounts, props | €30-50 |
Phase 1 POC total (Mini Talon + Lepton): ~€520-640 Phase 1 POC total (Skywalker X8 + Lepton): ~€570-720 Phase 2 operational (any airframe + Boson): add €600-950 to Phase 1
SOFTWARE SETUP¶
ArduPilot on H7A3-WING¶
- Flash ArduPilot Plane firmware via Mission Planner
- Configure airframe type (flying wing / conventional)
- Calibrate accelerometer, compass, RC
- Configure TECS (Total Energy Control System) for stable survey flight
- Set up failsafe: RTH on RC loss, RTH on low battery
Mission Planner (ground station, FOSS, runs on Linux via Wine or Windows)¶
- Alternatively: QGroundControl (better Linux support, also FOSS)
- Draw survey grid on map → auto-generates waypoints
- Upload to FC → arm → auto-takeoff → fly grid → auto-land
- Telemetry display: position, altitude, battery, airspeed, groundspeed
Survey Grid Planning¶
- Tool: Mission Planner Survey Grid tool or QGroundControl Survey
- Input: field boundary polygon, altitude, overlap percentage
- Output: waypoint mission file
- Typical settings: 60m altitude, 60% side overlap, 80% front overlap for photogrammetry-grade coverage (overkill for wildlife survey — 30% overlap sufficient)
Detection Processing¶
- Python script (to be written): reads GPS log + thermal images → KML + PDF report
- OpenCV for basic thermal blob detection (hot spot identification)
- QGIS for GIS overlay (FOSS, EU-friendly)
- No cloud dependency, runs on the van laptop
OPERATIONAL PROCEDURE¶
Before flight: 1. Check weather: wind <6m/s, no precipitation, temperature above FC minimum 2. Pre-flight checklist: battery, RC link, GPS lock, telemetry connected, payload powered 3. Upload survey grid mission to FC 4. Notify local airspace if required (Czech CAA DRONES app)
Launch: - Hand launch (X8, Mini Talon) or bungee launch for heavier airframes - Auto-takeoff mode: ArduPilot handles the climb to survey altitude
Survey: - Aircraft flies the grid autonomously - Operator monitors Mission Planner: position, battery, any anomalies - Ranger watches thermal feed (if HDZero installed) OR waits for post-flight data - Ranger presses trigger button on Radiomaster to manually flag detections
Recovery: - RTH (Return to Home) automatic or manual command - Auto-land or hand-catch depending on airframe and site
Post-flight: - Download SD card from payload ESP32-S3 - Run detection processing script - Generate PDF report + KML file - Deliver to myslivci contact
PAYLOAD INTERFACE NOTE¶
The thermal payload uses the same GX12-7 dual connector standard as libdrone Pro: - Connector A (left): signal — UART telemetry to FC, trigger input from RC - Connector B (right): power — 5V regulated from BEC
This means a thermal payload designed for libdrone Wing can plug into libdrone Pro backplane and vice versa. One payload library, two airframes. This is the core commercial argument for maintaining the standard across platforms.
RISKS¶
| Risk | Probability | Impact | Mitigation |
|---|---|---|---|
| ArduPilot tuning complexity | High | Medium | Large community, documented X8/Penguin builds. Expect 3-5 sessions to tune properly. |
| EASA regulatory burden | Medium | High | Aircraft >900g AUW in specific category needs operational authorisation. Budget time and possibly legal advice. Consider keeping AUW <900g with Mini Talon + Lepton. |
| Thermal resolution insufficient for report | Medium | High | Phase 1 Lepton may not produce professionally credible imagery. Manage myslivci expectations — POC data, not operational product. Boson for Phase 2. |
| Fixed-wing hand launch in field conditions | Medium | Medium | Practice launches in open field before mission deployment. X8 is forgiving. |
| ArduPilot auto-land accuracy | Medium | Medium | GPS accuracy sufficient for landing zone. Designate a clear landing area. Hand-catch as backup. |
| Range limitations of HDZero for long survey | Medium | Low | Use telemetry-only V2 approach — video link not required for autonomous survey. |
| Animal disturbance by aircraft noise | Low | Medium | Survey altitude 50-60m significantly reduces audible footprint vs quadcopter. Fixed-wing quieter in cruise. |
| Payload weight affecting flight characteristics | Low | Medium | Lepton is <50g — negligible. Boson at 80g needs careful CG management. |
OPEN POINTS¶
- OP1: Airframe selection — Mini Talon V2 vs Skywalker X8 for Phase 1. Decision driver: available budget and timeline.
- OP2: HDZero integration — is live thermal feed operationally necessary for the myslivci use case or is post-flight GPS log sufficient? Discuss with myslivci contact before building.
- OP3: EASA regulatory path for commercial survey flights over fields. Consult Czech CAA before first commercial deployment.
- OP4: Detection processing script — needs to be written. Python + OpenCV + reportlab PDF. Estimate 1-2 weekends.
- OP5: Myslivci pilot scope — define the exact first mission: which field, which spolek, what output format they need.
- OP6: GX12-7 payload mechanical integration into chosen airframe belly bay. Needs physical design work once airframe is selected.
NEXT ACTIONS (in priority order)¶
- Talk to the myslivci contact in Telč. What do they actually need? What format? What field? This conversation defines everything else.
- Select airframe based on budget available after libdrone Pro components are paid.
- Order FLIR Lepton 3.5 — long lead time item, order early.
- Flash and bench-test H7A3-WING with ArduPilot while waiting for airframe delivery.
- Write detection processing script — can be done on any evening, no hardware needed.
- First autonomous test flight — empty airframe, no payload, just proving ArduPilot works.
- Integrate thermal payload once airframe is flying reliably.
- First thermal test flight — fly over a known animal location (your own garden, a farm) to validate detection quality.
- First myslivci pilot mission — target October/November 2026, thermal window open.
THE PITCH TO MYSLIVCI¶
"We can fly your fields at dawn with a thermal camera, autonomously, and hand you a GPS map of every deer and wild pig we detected. One flight, one report, your evidence for the ministry. We want to do the first one for free to prove it works."
That's it. That's the whole pitch.
The free first flight is the investment. Everything after that is a service contract.
libdrone Wing V0.1 — March 2026 "Build it, they will come — but first, fly it."
Revision History¶
| Version | Date | Author | Summary |
|---|---|---|---|
| 0.1 | 2026-03 | JS | Initial Wing concept document. |