Problem
Detect what one sensor misses
The system explores whether a low-cost UAV payload can combine multiple physical signatures to improve detection coverage for UXO and landmine targets.
Investigation
Five sensing modalities
- Fluxgate magnetometer
- Passive thermal camera
- Laser speckle vibrometry
- SFCW GPR using NanoVNA-V2
- Hyperspectral imaging
Model
Physics + environmental context
The repository combines target physics models with environmental data and repeated fusion experiments. Sixteen fusion versions were explored; the documented final approach used peak extraction and spatial coincidence after Bayesian fusion showed probability-saturation problems.
Evidence
What the repository reports
The documented simulation campaign covered January and July 2024 conditions and reported 9/12 target detections with about 33% precision. These are simulation results, not a field-validation claim.
Research status: The project remains a research and development system. The next evidence step is an inert test lane, followed by hardware/algorithm refinement and field-oriented validation.
Repository