Livestock and fence checks

On this page 11

The daily round without the drive

Drones walk the fence line, count the herd and flag animals that are injured, missing or lying unusually still.

CategoryOperate
Sluglivestock-and-fences
CadenceDaily or twice daily through the grazing season
PayloadRadiometric thermal, zoom RGB camera, RTK
Needs a visit firstYes. A licence check and a route survey come first

The problem

Checking outlying grazing means a drive, a gate, a walk and an hour, every day, in all weather. Most days nothing is wrong. The days something is wrong, the delay between the fence going down and somebody noticing is the whole problem.

How it works

StepWhat happens
PatrolA fixed route covers the fence line and the grazing block on a schedule
CountAnimals are counted and compared against the number the block should hold
AssessThermal and visual signatures flag animals that are separated, lame, or lying still when the herd is grazing
ReportA short summary lands after each patrol, with photographs of anything that needs a human decision

What you get

  • Head count per block
  • Fence condition log
  • Flagged animal list with imagery
  • Grazing pressure map

What the dashboard measures

  • Head count against expected
  • Fence sections flagged
  • Animals flagged per patrol
  • Time from patrol to report

Where it matters most

Dairy and grazing, grassland, cooperatives.

Build it

This is the capability that most wants a dock, because its value is entirely in frequency and in not driving. A daily patrol that somebody has to drive to is the problem it was meant to solve.

The regulatory shape decides the build

A patrol over grazing land is usually beyond visual line of sight, over somebody's property, possibly near a road. In the EU that is the Specific category, which means an operational authorisation or a standard scenario. It is the reason this capability is marked needs a visit rather than being switchable on. See Regulation.

Build the visual line of sight version first. A patrol flown from the yard over blocks the pilot can see is legal today under the Open category with an A2 certificate, and it still removes the drive.

Hardware

ItemWhyEU sourceUS sourceIndicative
Airframe with 35+ minute enduranceFence lines are longHolybro X500 V2 or a fixed-wing VTOL for large platformsGetFPV€500 to €4,000
Zoom RGB cameraReading a fence at 40 m and a tag at 15 m are different jobsSony FCB block camera modulesB&H€600 to €2,500
ThermalSeparates an animal ruminating from an animal in trouble, and finds one in a hedgeInfiRay moduleFLIR Boson€900 to €4,000
LTE linkThe report has to leave the fieldTeltonika RUTX11SparkFun€150 to €250
Dock, if unattendedThe whole pointSee docks€3,000 DIY to €35,000 turnkey
Integrated alternativeZoom, thermal, RTK, dock support, out of the boxDJI Matrice 4T via SolectricAdvexure€9,000 to €15,000

In-house software

StageWhat we runLicenceReplaces
CountingDetection over nadir frames, deduplicated across overlapApache-2.0 modelscommercial livestock counting
Behaviour flagsPosture and position heuristics first: separated from the herd, lying while the herd grazes, not moved between passesown code
Fence inspectionChange detection along the fence line against the previous patrol, on a fixed routeown code
EscalationAnything ambiguous goes to a human reviewer before it becomes an alertthis repo
ReportOne short summary per patrol with imagery attachedthis repo

The escalation queue is not an admission of weakness, it is the product. A false "your cow is down" at 06:00 costs the customer's trust permanently. Route uncertain findings to a person and say so on the tin.

Fence inspection is genuinely hard from the air and works far better as a change detector than as a classifier. A fixed route makes "this 8 m section looks different from Tuesday" cheap and reliable, which is what a farmer actually wants.

The move is the sequel

Counting a mob and walking a fence line is the patrol. Actually moving stock between blocks is automated herding, which rides on this aircraft with one extra payload. Build the patrol first: it is the easier regulatory case, and it is what proves to a farmer that the counting is trustworthy before anything is asked to push their animals anywhere.

Cost efficiency

  • Visual line of sight first, dock second, BVLOS third. Each step multiplies the regulatory cost. Do not start at the end.
  • Counting is the anchor product. Head count against expected is unambiguous, verifiable and immediately valuable. Behaviour flagging is the upsell.
  • Thermal earns its place in winter and at dawn. In midday summer a black cow in sunlight is a thermal mess. Fly the patrol early.
  • Share the airframe with the wildlife search. Same thermal payload, same time of day, different season for the peak. One aircraft covers both.