Perimeter and border surveillance is a standoff game: detect early, verify without approaching, and hand a confirmed track to responders. A thermal channel finds intruders regardless of lighting; a co-boresighted zoom channel identifies them; laser ranging puts coordinates on the target. Coverage planning starts from detection distance and ends at the response time it buys.
- 1,042 mPerson detection, Johnson class
- 30xOP-125A optical zoom
- 2,000 mLaser ranging class
- 0.8 WLX-9B dual NIR illuminators
Key takeaways
- Plan sector length on the range at which you can verify, not detect. An unverifiable contact does not dispatch anyone.
- Published DRI figures use different standards that differ by an order of magnitude. Always check which one a number came from.
- Detection buys response time, and that is the metric to design around — how long a responder has between alert and arrival.
- Thermal is passive and NIR is invisible to the eye. Neither reveals the patrol’s timing the way a floodlight does.
On this page
Detection maths for perimeter work
Coverage planning starts from published DRI distances. Our long-range payloads are rated to detect a person at 1,042 m under Johnson criteria. That range defines tower spacing, orbit radius and how much fence one aircraft can watch.
Compare range figures only when the sensor channel, target size, task definition and test conditions are stated. Johnson-based thermal DRI and visible-camera DORI are not interchangeable, and there is no universal conversion factor between them. Use the model-specific source document and validate the required task under your operating conditions.
The practical planning rule is to design sector length on verification range rather than detection range, and to keep a working margin below whichever published figure you use.
| Target | Detection | Recognition | Verification / ID | Planning basis |
|---|---|---|---|---|
| Walking person | 1,000–1,800 m | 350–600 m | 150–350 m | Use verification |
| Crouched or prone person | 500–900 m | 200–400 m | 100–200 m | Use verification |
| Group of people | 1,200–2,000 m | 400–700 m | 200–400 m | Cluster reads early |
| Vehicle, engine warm | 2,500–4,000 m | 900–1,400 m | 400–800 m | Engine dominates signature |
| Vehicle, cold | 1,200–2,000 m | 600–900 m | 300–600 m | Much weaker signature |
| Small boat | 800–1,500 m | 350–600 m | 150–350 m | Depends on water temperature |
| Person in dense scrub | 300–600 m | 120–250 m | < 120 m | Occlusion dominates |
| Any target, rain or haze | Halve all figures | Halve | Halve | Weather margin |
Design around response time, not detection distance. A detection at 1,500 m is only useful if it gives a responder more time than a detection at 400 m would. Sector geometry, response posture and terrain access determine whether extra range converts into extra minutes.
Night verification without floodlights
Thermal detects but rarely identifies. Payloads pairing thermal with 30x optical zoom close that gap, and NIR laser illumination modules extend the zoom channel deep into unlit ground while staying invisible to the naked eye.
This combination is what makes covert perimeter watch practical. A floodlight identifies well and simultaneously tells the subject exactly where the patrol is looking and when. Thermal is entirely passive; NIR illumination is invisible to human vision, though detectable by other NIR-capable sensors — covert rather than undetectable.
Boresight alignment between the thermal and zoom channels matters more here than in most applications. The workflow is detect on thermal, slew, then identify on zoom, and if the two channels disagree about where they are pointing that hand-off fails at exactly the moment it matters. See EO/IR fusion and boresight alignment.
From contact to responder
A perimeter programme is judged on whether contacts become interceptions. That depends on the position hand-off far more than on the sensor.
Laser ranging computes a target coordinate directly from the observation, which converts a track into a dispatch. Report it with an accuracy radius rather than as a point — angular error grows with slant range, and perimeter observation is usually oblique rather than near-nadir. On a 1,500 m observation, a one-degree heading error alone contributes about 26 m of lateral error, which the terms in the geolocation error budget break down.
AI multi-object tracking earns its keep where one operator watches long sectors: automatic flagging turns continuous scanning into alert monitoring, and click-to-track holds the subject while the aircraft repositions for a better angle.
What perimeter programmes get wrong
- Planning on detection range. Produces contacts that cannot be classified and a team chasing warm animals.
- Mixing DRI standards between vendors. Compare like with like or the comparison is meaningless.
- Ignoring weather margin. Rain and haze roughly halve every published figure, and perimeters do not get to choose their weather.
- Neglecting boresight. Thermal-to-zoom hand-off fails silently when alignment drifts after a hard landing or thermal cycling.
- No retention and access policy. Perimeter imagery contains people, staff and neighbouring property; decide the policy before the first complaint rather than after.
- Treating the drone as a replacement for fixed sensing. The efficient architecture is usually fixed sensors on high-risk sectors and the aircraft as the response and coverage asset.
Recommended UAVThermal payloads
OP-125A — flagship EO/IR payload in the 1,042 m person-detection class, with 30x optical zoom, 2,000 m laser ranging and AI tracking. The standard choice for long sectors.
LX-9B — 640×512 thermal with 30x zoom and dual 0.8 W NIR laser illumination for verification on unlit ground.
LX-8RC — long-reach hybrid zoom with laser night vision for extended fence lines where standoff dominates the requirement.
Related reading
Technology: how DRI figures are calculated and NIR laser illumination.
Field practice: DRI ranges explained, real detection data from six payloads and AI target tracking.
Adjacent missions: night security patrol and port and harbour surveillance.
FAQ
How far can a drone realistically detect a person at night?
Spec-sheet detection for 640-class payloads in this family is 1,042 m under Johnson criteria. Plan operations around a working margin below that, because rain, haze and humidity roughly halve every published figure and because detection is not the same as being able to classify what you found.
Why do published detection ranges vary so much between vendors?
Compare range figures only when the sensor channel, target size, task definition and test conditions are stated. Johnson-based thermal DRI and visible-camera DORI are not interchangeable, and there is no universal conversion factor between them. Use the model-specific source document and validate the required task under your operating conditions.
Does the patrol reveal itself at night?
Not through the sensors. Thermal is fully passive and emits nothing, and NIR illumination is invisible to human vision — covert, though detectable by other NIR-capable sensors. A visible floodlight, by contrast, tells the subject exactly where and when you are looking.
Should a drone replace fixed perimeter cameras?
Usually not. The efficient architecture is fixed sensing on high-risk sectors, with the aircraft covering long low-risk stretches where cabling cost is prohibitive and responding when a fixed sensor alarms.