Counter-UAS Sensor Fusion: Resolving Sensor Conflicts

Learn how counter-UAS systems resolve radar and optical sensor conflicts. Explore timing, calibration, uncertainty weighting, and infrared seeker field of view for reliable counter-drone systems tracking.

10/29/20262 min read

Infographic explaining sensor fusion for conflict resolution combining radar and optics data into a fused track.
Infographic explaining sensor fusion for conflict resolution combining radar and optics data into a fused track.

1. They measure different things
Radar gives range and speed, plus angle with an error of several to tens of mrad, updated slowly. Optics gives a sharp angle and shape, but no range, updated fast. There is no single “truth.” There are two views of the world, in two clocks and two coordinate frames.

2. Averaging is the rookie mistake
The instinct is to average the two tracks. Wrong. The average of two disagreeing tracks sits between them - in empty air. Fusion is not averaging. It is conflict resolution: pick the source that is reliable for the quantity you need, right now.

3. Cue error vs seeker FOV
Radar cues the optics where to look. But if the radar angular error is bigger than the seeker’s narrow FOV, the ROI points past the target. So the first lock needs a wide FOV or a few cue positions (this is Part 3 again). A tight gate fed by a loose cue finds nothing.

4. Time alignment, not fresh data.
Radar ticks at a few Hz. Optics at 30-60 Hz. A “fresh” radar blip and a “stale” optics frame describe different moments. At 350 m/s closing speed, that gap is meters. Timestamps must be aligned to one clock before any fusion, or the math fuses two different times and lies.

5. One coordinate frame, calibrated
Radar lives in its mount frame, optics in its own, the airframe bends and shifts. Without a shared inertial/world frame and a checked extrinsic calibration, the two sensors report the same target at two angles. The conflict is sometimes just a misalignment you never measured.

6. Weight by innovation, not by habit.
Each sensor carries its own uncertainty. When a measurement jumps far from the predicted state (large innovation), that sensor’s weight drops for this cycle. Radar owns range and speed. Optics owns angle and shape. Let each lead where it is strong, and trust neither where it is weak.

7. Degrade, don't break.
Rain and fog kill the optics; radar holds. Jamming and clutter fool the radar; optics with shape holds. A good fusion stack does not crash when one channel dies. It re-weights and keeps tracking on the survivor. Silence from one sensor is information, not failure.

Bottom line:
Sensor conflict is normal. The architecture must expect it:
- never average two disagreeing tracks;
- match cue error to seeker FOV;
- align time before aligning space;
- share one calibrated frame;
- weight by innovation per quantity;
- degrade gracefully when a channel goes blind.

Next Part 7: how do you prove any of this works without firing the interceptor? Validation, simulation and the metrics that actually matter.

At SpearX, building uncooled LWIR guidance for CUAS, I learned that fusion is not “add more sensors.” Fusion is deciding who to believe when they disagree.

Related : GEOCOM Co. LLC www.geocomco.eu DeepTechRnD www.deeptechrnd.eu SpearXAgro www.spearxagro.eu

Contacts

SpearX ↑ Strike with Precision

+359 877 620 210

D-U-N-S® 525540791

© 2025 - 2026. All rights reserved. SpearX Projects. Geocom Co. LLC

SpearX is an EU-Ukrainian company located in a strategic areas for R&D in UAV