DIY Trinity Pro Equivalent — Cebu Flood Resilience AI | TalaStar
Open Source Cebu Pilot v1.0

DIY TRINITY PRO
EQUIVALENT

Building a flood-resilience UAV for Cebu, Philippines using open-source components. Autonomous aerial intelligence at a fraction of commercial cost.

View BOM AI Pipeline
$7,580
Total Build Cost
72 km²
Coverage / Mission
12 weeks
MVP Build Time

01 — REFERENCE SYSTEMS

Commercial benchmarks that define what our DIY build must match or exceed.

Primary Reference

TRINITY PRO

Wingtra — $30,000–$50,000

TypeFixed-wing VTOL
Endurance59 min
Coverage72 km²/mission
Payload600 g
Accuracy (GSD)1.5 cm/px
PPK / RTKYes — survey-grade
Target marketSurveying, mapping, precision agriculture
LiDAR Benchmark

QUBE 240 LiDAR

YellowScan — $22,000–$30,000

TypeSolid-state LiDAR sensor
Range240 m
Accuracy±2 cm
Weight630 g
Points/m²100–300
ReturnsUp to 4 returns (first, last, inter)
Best forCanopy-penetrating terrain surveys, flood modelling
RGB Payload

SONY RX1R II

Full-frame 42MP compact — $3,000

Sensor42.4 MP full-frame CMOS
Focal length35mm f/2
ISO rangeISO 50–102400
VideoFull HD 60fps
Weight507 g (body only)
GSD target2 cm/px @ 120m AGL
HazardNo stabilisation; requires gimbal

02 — DIY ENGINEERING BLUEPRINT

Open-source architecture selected for each subsystem. All components verified against Trinity Pro equivalent performance targets.

AIRFRAME #1 subsystem
Platform
Multirotor with VTOL conversion
Recommended Frame
DJI M600 Pro (refurbished)
Frame Weight
~4.8 kg (with battery)
Max Takeoff Weight
15.5 kg
Payload Capacity
1.5 kg reserved for sensors
Budget Alternative
Tarot 920mm hex frame ~$280
AUTOPILOT — PIXHAWK / ARDUPILOT #2 subsystem
Primary flight controllerPixhawk 2.4.8
FirmwareArduPilot Copter 4.x (stable)
IMU redundancyDual IMU with temperature compensation
GPS / RTKHereFlow RTK GPS + u-blox ZED-F9P
Ground stationMission Planner / QGroundControl
TelemetrySiK 915 MHz radio, 2 km range
Why Pixhawk over DJI N3?
Open-source means full parameter control for autonomous survey missions. DJI's SDK is locked — you cannot run custom AI inference on the flight controller. Pixhawk + companion computer (Raspberry Pi 4 / Jetson Nano) gives you the full pipeline: real-time flight control + edge inference on a single platform.
LIDAR — LIVOX AVIA Phase 2 upgrade
Range190 m @ 80% reflectivity
Points/sec240,000
FOV70.4° × 70.4°
Weight (body)480 g
Power draw12 W typical
ResistanceIP66 — rain/flood operations
Cost~$2,800 (vs. Qube 240 $22k+)
DIY advantage: Livox Avia (~$2,800) delivers within 4× the point density of Qube 240 at 1/8th the price. Post-processed in CloudCompare or OpenCV, it produces terrain models sufficient for flood-risk mapping in Cebu's urban river catchments.
RGB CAMERA PAYLOAD Phase 1 MVP sensor
Camera
Sony α6000 body + Sigma 16mm f/1.4
Resolution
24.2 MP APS-C
GSD at 120m
2.8 cm/px
Trigger
Miranda + Chasing-PP
Gimbal
Gremsy T3 (3-axis stab)
Total cost
~$1,400
MULTISPECTRAL RIG Phase 3 upgrade

NDVI and NDWI analysis for flood-plain moisture mapping. Bands selected for Cebu's vegetation index calibration cycle (Mar–Oct).

BLUE
450 nm
Water absorption
GREEN
550 nm
Vegetation peak
RED
660 nm
Chlorophyll abs.
NIR
850 nm
Cell structure ref.
REDEDGE
735 nm
NDWI edge
Recommended sensor: MicaSense RedEdge-M (5-band, $3,200) or DJI P4 Multispectral. Budget alternative: ArduCam multi-filter stack on Raspberry Pi HQ camera — $400 all-in for the band array, requires custom calibration.
POWER SYSTEM #5 subsystem
Battery
6S 22,000mAh LiPo (2× parallel)
Flight time (RGB)
42 min
Flight time (+LiDAR)
28 min
ESC
40A × 6 (OneDrive 4-in-1)
Power budget
~800W peak draw
SOFTWARE STACK Full open-source
Flight & Mission
AutopilotArduPilot Copter 4.x
Ground ControlMission Planner / QGC
Survey PlanningAutomated grid (ArduPilot spline)
GeotaggingExifTool + GPS sync log
Processing & AI
Imagery / SfMOpenDroneMap (ODM), WebODM
Point cloudsCloudCompare, PDAL
Flood AIPython + TensorFlow Lite (edge)
GIS outputQGIS, GeoJSON, PostGIS
All processing runs on a companion Raspberry Pi 4 (4GB) for mission-logging. Full AI inference (flood detection, damage classification) runs on a Jetson Nano at the base station — not on the UAV — to stay within power/weight budget.

03 — BILL OF MATERIALS

All-in Phase 3 build cost. DIY saves 80–90% versus commercial equivalent systems.

# Component Category Phase Unit Cost (USD) Qty Subtotal Source
1DJI M600 Pro frame (refurbished)AirframeP1$1,8501$1,850DJI refurb / China
2Pixhawk 2.4.8 + IMUAutopilotP1$1801$180Holybro / China
3u-blox ZED-F9P RTK GPSAutopilotP1$2401$240CRS Belgium
4SiK 915 MHz telemetry radioAutopilotP1$552$110HK / China
5Sony α6000 + Sigma 16mm f/1.4CameraP1$9001$900Japan / PH
6Gremsy T3 gimbalCameraP1$4801$480Gremsy / Taiwan
76S 22,000mAh LiPo (2× + charger)PowerP1$3202$640CN / PH hobby
840A ESC × 6PowerP1$456$270China
9Raspberry Pi 4 (4GB) + caseComputingP1$801$80RS / PH
10ODM workstation (mini-ITX build)ComputingP1$6001$600DIY
11Mission Planner laptop (used ThinkPad)ComputingP1$2501$250PH used market
12Livox Avia LiDARLiDARP2$2,8001$2,800Livox direct / HK
13Livox Avia protective housingLiDARP2$1801$180DIY / 3D printed
14Jetson Nano (4GB) + carrier boardComputingP2$3501$350NVIDIA dev kit
15MicaSense RedEdge-MMultispectralP3$3,2001$3,200MicaSense / US
16Sunshine sensor + calibration panelMultispectralP3$4801$480MicaSense
17Spare parts & contingency (15%)OtherAll$1,137$1,137
TOTAL BUILD COST (Phase 1 + 2 + 3 + contingency) $13,747
Phase 1 MVP only $5,600
$5,600
Phase 1 MVP
12-week build
$9,930
Phase 1 + 2
+ LiDAR, 20 weeks
$13,747
Full System
+ Multispectral, 30 weeks
$36,000+
Commercial equivalent
Trinity Pro + Qube 240 + RX1R II

04 — PHASED BUILD STRATEGY

Incremental capability building. Each phase produces deployable outputs, not dead-end prototypes.

PHASE 1 Weeks 1–12

RGB SURVEY MVP

Deploy a basic aerial survey capability first. Validate community trust and data quality before committing to LiDAR investment.
Deliverables
  • ✓ RGB orthomosaic of priority flood zones
  • ✓ DSM/DTM from structure-from-motion
  • ✓ Community trust infrastructure in place
  • ✓ ArduPilot survey grid automation
  • ✓ Pilot training completed (2 operators)
Budget $5,600
PHASE 2 Weeks 13–20

LIDAR PENETRATION

Add canopy-penetrating LiDAR for terrain modelling beneath dense vegetation in Cebu's upstream catchments.
Deliverables
  • ✓ Point cloud terrain model (2 cm accuracy)
  • ✓ Watershed elevation profile
  • ✓ First-flush flood modelling data
  • ✓ AI edge inference pipeline (Jetson Nano)
  • ✓ CAAP BVLOS waiver application draft
Budget (cumulative) $9,930
PHASE 3 Weeks 21–30

FULL AI RIG

Deploy the complete multispectral + AI pipeline for real-time flood severity scoring and community alerting.
Deliverables
  • ✓ 5-band NDVI/NDWI flood mapping
  • ✓ Real-time damage scoring AI (< 30 min post-flight)
  • ✓ Community SMS/chat alert integration
  • ✓ BVLOS operational certification
  • ✓ Open dataset publication (Cebu Flood Dataset v1.0)
Total cost $13,747

05 — RESEARCH GAPS

Seven open problems that prevent this system from achieving commercial-grade performance. Each is a publishable research contribution.

01

Cebu Dataset Gap

No labelled UAV dataset exists for Cebu's flood topography. Training data must be generated from scratch, validated against PH-EWARS benchmarks, and open-sourced to support repeatability.

Data Acquisition
02

VTOL Typhoon Performance

No published data on multirotor flight envelope during Cebu's typhoon seasons (Nov–Jan, 140+ km/h gusts). Autonomy fails without a published gust-recovery protocol and airframe stress modelling for sustained exposure.

Flight Safety
03

LiDAR + Multispectral Fusion

No established methodology for fusing Livox Avia point clouds (phase-shifted, range-dependent density) with MicaSense 5-band imagery for flood-severity scoring. Requires novel co-registration and feature fusion pipeline.

Algorithm
04

Edge AI Deployment

Jetson Nano inference throughput for real-time flood segmentation (U-Net variant) is unverified on tropical urban imagery. Quantisation and model compression required; current literature is temperate-climate biased.

Edge ML
05

Community Scoring Framework

No validated community flood-risk scoring framework exists for Cebu's barangay structure. Metrics must be co-designed with local government units (LGUs) — model cannot be top-down imposed if it is to be trusted and sustained.

Governance
06

CAAP BVLOS Regulations

Philippine CAAP Civil Aviation Regulation Part 107 does not yet provide clear BVLOS pathways for community UAV operations. Research into the regulatory gap, international ICAO comparators, and a proposed framework for PH is needed.

Regulatory
07

Cost-Effectiveness Study

No published cost-per-km² analysis comparing DIY UAV flood monitoring vs. satellite imagery (Sentinel-2) vs. manual survey for Cebu's urban-rural mix. Essential for securing LGU and donor funding for Phase 2–3 expansion.

Economics

06 — FLOOD AI PIPELINE

End-to-end data flow from aerial capture to actionable community alerts. Built with open-source tooling throughout.

INPUT
RGB / LiDAR / Multispectral
Aerial capture
PRE-PROCESS
ODM → GeoTIFF
Stitching + georef
ANALYSE
U-Net Segmentation
Flood extent detection
ENRICH
NDWI + Terrain
Severity scoring
DELIVER
GeoJSON + Alerts
LGU + community
Data Inputs
RGB ORTHOMOSAIC
2.8 cm/px at 120m AGL. Used for change detection and structural damage classification.
LIDAR POINT CLOUD
240k pts/sec. Ground returns used to generate terrain model beneath canopy. ±2cm vertical accuracy.
MULTISPECTRAL BANDS
Blue/Green/Red/NIR/RedEdge. NDWI for standing water, NDVI for vegetation saturation.
HISTORICAL BASELINE
Sentinel-2 archive (ESA) for seasonal change comparison. 10m resolution, free access.
Output Products
FLOOD EXTENT MAP
GeoJSON polygon of inundation area. Used by LGU for evacuation zone definition.
DAMAGE SEVERITY INDEX
1–5 scale per barangay. Combines inundation depth, structural exposure, and population density.
TERRAIN RISK SCORE
Slope + drainage + soil-type composite. Updated seasonally from LiDAR data.
COMMUNITY ALERT
SMS/chat message per barangay. Triggered when damage index exceeds threshold. < 30 min from mission end.

07 — PROCUREMENT SOURCES

Recommended sourcing by region. All prices indicative and subject to market movement.

🇨🇳 CHINA
Primary cost-efficiency source
  • • Livox direct (Shenzhen) — LiDAR
  • • Banggood / AliExpress — ESCs, frame parts
  • • Holybro (via China warehouse) — Pixhawk
  • • Tarot — airframe alternatives
Lead time 4–6 weeks. Factor customs + PH import duties (~30%).
🇵🇭 PHILIPPINES
Local support + fast delivery
  • • Amax Enterprise — DJI (Manila)
  • • Hobby Warehouse PH — batteries, parts
  • • PC Express / Datablitz — computing (RPi)
  • • Used ThinkPad market (Facebook PH)
Refurbished M600 Pro ~$1,850 via PH DJI dealer. In-country warranty.
🇹🇼 TAIWAN
Quality gimbal + industrial
  • • Gremsy (Taoyuan) — gimbal direct
  • • RTS Technology — industrial computing
  • • Taiwan PCB shops — custom PCB for integration
Gremsy ships to PH in 5–7 days. Technical support responsive via email.
🌐 INTERNATIONAL
Specialised / high-precision
  • • CRS Belgium — u-blox ZED-F9P RTK
  • • MicaSense (USA) — multispectral sensors
  • • NVIDIA — Jetson Nano dev kit
  • • RS Components — Raspberry Pi (Manila)
MicaSense requires export licence for some jurisdictions. Budget 8 weeks for delivery + customs.
↑ Back to top