Define the Mission Before You Choose a Single Component

"A drone that patrols the site" is not a specification, and teams that start there end up with an aircraft nobody buys. The questions that actually determine the hardware are narrow and answerable: How large is the perimeter in linear feet? How many rounds per night? Is the drone doing scheduled patrols, responding to alarms from fixed sensors, or both? Does it need to identify a person at 200 ft in darkness, or only confirm that a fence line is intact? Is a human watching the feed live, or is the aircraft expected to triage on its own and escalate?

Those answers set everything. A 40-acre logistics yard with a 4,000 ft perimeter and four rounds per shift is a 12 to 18 minute mission repeated eight times, which drives endurance, charge time, and dock count. A refinery with a hazardous-area classification and a requirement to read gauge faces is a different aircraft entirely. Write the mission profile as a timed sequence before anyone opens CAD.

The Docking Station Is the Actual Product

Customers do not buy an aircraft. They buy uptime with no person on site, and the thing that delivers uptime is the dock. In a well-designed system the aircraft is the cheaper and simpler half.

A production dock has to handle precision landing to within an inch or two, which usually means a fiducial marker plus vision guidance and a passive centering funnel or a set of powered arms that mechanically square the aircraft after touchdown. It has to make charge contact reliably through dust, ice, and pollen, which favors sprung contacts with a wiping action over flat pads. It has to protect the aircraft from weather while closed and open under load, which means a lid mechanism rated for tens of thousands of cycles with heaters for ice and drains for rain. It needs internal climate control, because a LiPo pack charged at 20°F degrades quickly and one stored at 120°F degrades faster. And it needs to recover on its own when something goes wrong, because a truck roll to a remote site costs more than the flight it saves. The full architecture is worked through in drone-in-a-box docking stations.

Charge time sets the duty cycle. A pack that flies 20 minutes and charges in 40 gives roughly a 33 percent availability figure per aircraft. If the customer wants continuous coverage, the answer is two aircraft per dock or two docks, and that arithmetic belongs in the proposal rather than in a surprise conversation after installation.

Payload, Autonomy, and the Link

The default payload is a dual-sensor gimbal: a visible camera with 20x to 30x optical zoom for reading plates and faces, and a thermal sensor for detecting warm bodies in darkness and through light foliage. Thermal is what makes the night patrol work at all, and its detection range limits should be measured on your actual site rather than taken from a datasheet. Payload selection is covered in thermal camera payload integration.

A spotlight and a two-way speaker convert detection into deterrence. In practice, an audible challenge from an aircraft resolves most trespass incidents without a guard ever leaving the gatehouse, and customers value that far more than image quality.

Autonomy has to cover the failure cases, not just the happy path. The aircraft must fly a repeatable route in wind, avoid a crane or a parked trailer that was not there yesterday, and return to a safe state on link loss, low battery, or GPS degradation near steel structures. Sensor and algorithm options are laid out in drone obstacle avoidance.

Communications need two independent paths in most deployments: a site Wi-Fi or private LTE link for video, and a lower-bandwidth cellular or radio channel for command and telemetry that survives when the video link drops. Video latency above roughly 300 ms makes live manual takeover unusable, so measure it end to end rather than assuming.

Reliability, Maintenance, and Getting to a Series

A patrol drone flying eight sorties a night accumulates 2,000 to 3,000 flights per year, which is one to two orders of magnitude more cycles than most drone products ever see. That changes what matters:

  • Wear items become the cost driver. Propellers, motor bearings, gimbal slip rings, and dock lid actuators all have finite lives. Establish replacement intervals through accelerated testing before you sell a service contract.
  • Battery lifecycle is a consumable. At 300 to 500 useful cycles, a pack lasts a few months on this duty cycle. Price replacement into the contract and design the aircraft for tool-free pack swaps.
  • Remote diagnostics are mandatory. Every flight should log health telemetry to a fleet backend that flags a degrading motor before it fails. The operational model is the same one described in IoT fleet management.
  • Design for field service by a non-specialist. The person who visits the site is a security technician, not an aerospace engineer. Modular assemblies, keyed connectors, and a guided replacement procedure decide whether the product is supportable.

Regulation, Noise, and Fitting Into the Existing System

In the United States, autonomous patrols without a visual observer require operating beyond visual line of sight, which means a waiver supported by evidence about the aircraft, the procedures, and the site. Plan for it early, because the airworthiness documentation it demands shapes the design. The path is described in BVLOS operations and FAA waivers, and every unit must comply with Remote ID requirements.

Noise kills deployments near residential property. A multirotor at 150 ft is clearly audible, and a system that runs eight times a night will generate complaints. Larger, slower propellers and higher patrol altitudes both help, and the acoustic signature should be measured and disclosed rather than discovered by the customer.

Finally, the drone must join a security operation that already exists. That means feeding the same video management system the fixed cameras use, accepting alarm triggers from the access control and fence sensor systems, and producing patrol logs in a format the customer's compliance process accepts. A drone that requires its own separate console is a second job for the guard, and guards will stop using it.

Scope the System, Not Just the Aircraft

Projects House develops autonomous security UAS as complete systems: aircraft, dock, autonomy, fleet backend, and the integrations the customer's existing security stack requires. Describe your site profile and coverage target through our contact form and we will scope it.