SenzFusion
Sensor-fusion payload for UAVs
A sensor-fusion payload that navigates the host UAV without GNSS, and detects, recognizes and tracks objects. Radar map matching corrects drift, a thermal scene check confirms the fix, and GNSS input is used to detect a contested environment.
- Sheet
- SenzFusion · Module 01
- Layer
- L1 · Fusion kernel
- Status
- On the aircraft · Payload + SDK
- Revised
How SenzFusion estimates position
Between fixes, the IMU carries the estimate. A position fix is accepted only when the radar and thermal matches agree.
- 01Primary · continuousPredict motionThe IMU predicts position, velocity and orientation 1,000 times a second. Barometer and airspeed feed the same fusion core.
- 02Ground profileRadar altimeterGround heights along the track build the terrain profile beneath the aircraft.
- 03Drift correctionRadar map matchThe radar profile and imaging-radar ground patch are matched to the stored terrain map.
- 04Cross-checkThermal scene checkA thermal terrain picture is checked against its reference scene.
- 05Position fixRadar + thermal agreeThe estimate is updated along-track and cross-track when both matches agree.
- 06Course correctionHost autopilotReceives the position estimate over a signed serial link and retains flight control.
- —GNSS · check onlyDetects contested environmentsUses GNSS input to detect a contested environment: satellite position is compared against the fused estimate.
Navigation responsibilities
- IMU
- Motion prediction between fixes
- Radar altimeter
- Ground profile for terrain map matching
- Imaging radar
- 2D ground patch
- Thermal camera
- Terrain picture for scene checks
- Barometer and airspeed
- Supporting inputs
- GNSS · check only
- Detects a contested environment
- FPGA system-on-chip
- Fixed-timing logic and ARM cores
- Host autopilot
- Keeps flight control throughout
A position fix is accepted only when the radar and thermal matches agree; between fixes, the IMU carries the estimate.
How navigation worksDetect, recognize and track an object
The Vision SDK streams video from the payload. Edge ATD/ATR is coming soon.
Terminal Guidance System (TGS)
Automatic Target Detection (ATD) and Automatic Target Recognition (ATR) for fire-and-forget applications.
Human in the loop
Target decisions stay with the operator and the host system.
Two jobs, one payload
Terrain matching locates the aircraft; target tracking follows the object.
Video and tracks go to OspreyVMS for operator review. Whether navigation and tracking share one thermal camera is decided per integration.
Edge computing on the vehicle
An FPGA system-on-chip, with fixed-timing logic and ARM cores, processes sensor data on the vehicle and matches it against references prepared before flight.
Live inputs
- IMU: motion prediction
- Barometer and airspeed
- Radar altimeter: ground profile
- Imaging radar: 2D ground patch
- Thermal camera: terrain picture and target imagery
- GNSS: check only — detects a contested environment
Processing core
- Time-stamp and clean sensor readings
- Predict motion from the IMU
- Build the ground profile
- Radar map match: profile vs terrain map
- Thermal scene check: picture vs reference scene
- Update the estimate along-track and cross-track
Estimate, fix and health
- Position estimate with uncertainty
- Course correction → host autopilot over signed serial
- Navigation confidence and payload health → SquadSense (coming soon)
- Video and object tracks → Vision SDK and mission applications
Holds terrain maps, thermal reference scenes and mission configuration for the pipeline.
The flight plan is delivered to the host separately from the payload’s reference data.
Proposed design, not a board specification. Storage, power, heat, sensor interfaces and latency still need validation on each platform. Timing and operating envelope not yet validated.
Where SenzFusion sits among the ShoalOS modules
Integrate SenzFusion into your aircraft
Define the sensor interfaces, payload fit and autopilot connection for your platform.