Report 102 · Luxury Home Security
Autonomous perimeter drones are now a standard line item in estate security proposals. The aircraft is the easy part. The hard part is a federal rule that says a person has to be able to see the thing with their own eyes for the entire flight, and a second rule that makes your covert overwatch asset announce its own position, and its operator's position, once per second.
I spent my Army career on the other side of this problem, working out what an unmanned system emits and what you can do about it. So when the estate security market started selling drone-in-a-box perimeter patrol, two questions were reflexive for me, and neither of them is about the drone.
Who is legally flying it, and what is it transmitting?
Both answers are public, both are in federal regulation, and neither one shows up in the brochure. This is the same failure mode I wrote about in why more gear will not protect your estate: the capability is real, the constraint on the capability is real, and only one of them is in the proposal.
Commercial drone flight in the United States runs under 14 CFR Part 107. The relevant clause is 107.31, and it is short enough to take seriously.
It requires that the remote pilot in command, the person manipulating the controls, or a visual observer be able to see the unmanned aircraft throughout the entire flight, with vision unaided by any device other than corrective lenses. That ability has to be sufficient to know the aircraft's location, determine its attitude, altitude and direction of flight, observe the airspace for other traffic or hazards, and determine that the aircraft is not endangering the life or property of another.
Unaided vision. Throughout the entire flight. A camera feed does not satisfy it, which is the point people miss. Your drone can be streaming 4K thermal to a monitoring centre in another state and still be in violation, because the regulation is about a human eye on the aircraft, not a human eye on the picture.
Put that against the marketing and the shape of the gap is obvious. A drone that launches itself on a randomised schedule at 3am and flies a perimeter route needs, under baseline Part 107, somebody standing outside in the dark watching it go.
I want to be fair here, because this is not a story about deception. Sunflower Labs makes one of the better-known residential systems, the Bee and its Beehive dock. Their own site describes the aircraft launching in seconds automatically via motion sensors or cameras, on a randomised schedule, or manually on demand, with a maximum operational radius of about 600 m.
And then, in the same body of text, this:
While the system is fully autonomous, most countries require that you have a pilot-in-command (PiC) connected to the system when the Bee is flying.
That is a vendor telling you the truth in plain language. The autonomy is real. The autonomy does not remove the human. Those two facts coexist, and the second one is a staffing line, not a technical footnote.
The layer that loses this is not usually the manufacturer. It is the estate security explainer content sitting above them, where "drones patrol large estates with 360-degree thermal visibility" appears with no mention that a certificated remote pilot is attached to the operation.
By getting permission. 107.31 is on the list of sections that can be waived under 107.205, alongside the visual observer rule, operation over people, the multiple-aircraft rule, and the general operating limitations. The FAA can waive visual line of sight if you demonstrate your operation is safe without it.
That waiver, not the airframe, is the product.
Sunflower Labs announced exactly this on 4 November 2025: a nationwide performance-based Certificate of Waiver or Authorization. Their release says it lets customers "legally operate BVLOS flights, fly over people and vehicles, and operate in low-visibility conditions, without requiring case-by-case approval," across 99% of the U.S., and describes the framework as one that anticipates the FAA's proposed Part 108 rules for routine BVLOS operations. Co-founder and CEO Alex Pachikov:
Through extensive collaboration with the FAA, we've secured authorizations that let us operate safely and legally across nearly every location in the United States.
Read that carefully, because it is the single most useful sentence in this report. The authorisation belongs to the operator, not to you and not to your house. If you buy that company's system and it is flown under their certificate, you are inside their envelope. If you buy a dock and a drone from somebody else and have your own staff fly it, you are back in baseline Part 107 with a person outside watching, unless you go get your own waiver.
"Is this system approved for BVLOS" is therefore the wrong question. The right one is: whose authorisation is this flight operating under, and may I see it?
The rule that would normalise all of this is real, and it is not finished.
The FAA published "Normalizing Unmanned Aircraft Systems Beyond Visual Line of Sight Operations" as a proposed rule on 7 August 2025, at 90 FR 38212. It runs 180 pages. Comments closed 6 October 2025, then the agency reopened the comment period twice, at 91 FR 3695 on 28 January 2026 and again at 91 FR 5880 on 10 February 2026. As of today it is still a proposed rule. Searching the Federal Register's own record for FAA documents under that title returns no final rule.
The abstract is worth reading for what it names as the target uses:
This proposed rule is intended to provide a predictable and clear pathway for safe, routine, and scalable UAS operations that include package delivery, agriculture, aerial surveying, civic interest, operations training, demonstration, recreation, and flight testing.
Private residential security patrol is not on that list. I would not over-read that, since the categories are broad and the final rule may land differently. But if you are being told your estate drone programme is about to become routine because Part 108 is coming, it is worth knowing that the proposal's own statement of purpose does not name your use case, and that the rule has not been issued.
Here is where my old job starts talking.
A drone that must be registered has to comply with Remote ID under 14 CFR Part 89. A standard Remote ID aircraft broadcasts a defined set of message elements, and 89.305 lists them:
Under 89.310 the aircraft may not take off unless it is broadcasting those elements, and the broadcast rate is at least one message per second.
Now think about that as a signature rather than as a compliance item. Your perimeter security asset transmits, in the clear, on a schedule, its own position and the position of the console flying it. Anyone within receiver range with a consumer phone app gets a live feed of when your patrol is airborne, where it is, how fast it is moving, and roughly where your operator sits.
For the regulatory purpose Remote ID was built for, accountability in shared airspace, that is exactly right and I have no argument with it. For a security asset it inverts a basic principle. I have written before about how cheaply a wireless camera link can be taken off the air. This is the companion problem, and it is arguably worse, because there is nothing to defeat. The aircraft is required to announce itself, and announcing itself is a condition of being allowed to fly.
A patrol that is observable is a patrol that is schedulable by somebody else. Randomised launch timing, which Sunflower explicitly offers, is a sensible countermeasure to exactly this, and it tells you the people building these systems have thought about it. It does not change the fact that once the aircraft is up, it is talking.
Over people. 107.39 prohibits operating a small unmanned aircraft over a human being unless that person is participating in the operation, is under a covered structure or inside a stationary vehicle offering reasonable protection from a falling aircraft, or the operation meets one of the operational categories in Subpart D. Your own guests, your own staff, your own family on your own lawn are human beings under this rule. A perimeter route that crosses the terrace during a party is a regulatory question, not just a courtesy one.
Altitude and the wall problem. 107.51 caps the aircraft at 400 feet above ground level, with an exception permitting higher flight within a 400-foot radius of a structure so long as it stays within 400 feet of that structure's uppermost limit. It also sets a minimum flight visibility of three statute miles as observed from the control station location, plus cloud clearance of 500 feet below and 2,000 feet horizontally. Fog and low cloud are weather conditions your fence does not care about and your drone does.
Everything that is not the FAA. Federal aviation law governs the aircraft. It does not settle state privacy statutes, local ordinances, trespass and nuisance law, or what happens when the camera on a patrol route sees over a neighbour's wall. I have not researched that layer for this report and I am not going to pretend otherwise, but on a dense luxury street it is likely the layer that actually generates your first complaint.
I help design the AI security systems for a veteran-owned (SDVOSB) luxury home-security company run by fellow veterans; I do not own it and earn nothing from this link. Full policy here. That list is close to what I ask when an aerial component is proposed for a property, and you should know I have a hand in that work when you read me recommending it.
I did not read the Sunflower Labs Certificate of Waiver or Authorization itself. Its scope is described here as the company described it in its own announcement, verified against two independent outlets that carried the release, and a waiver's real limitations live in the document rather than the press release. I also could not open the FAA's own Remote ID page, which returns 403 to automated retrieval, so every Remote ID statement above is taken from the regulation text in 14 CFR Part 89 rather than from FAA guidance.
I have not verified how many BVLOS waivers exist for private security work, or how many estate programmes are running under one versus running with a visual observer versus running in quiet non-compliance. I looked, and the numbers I found came from vendor marketing rather than FAA data, so I have left them out.
Part 108's status is stated as of publication. It is a rule in active motion and this paragraph is the one most likely to need a dated correction.
The pitch for an autonomous security drone is that it removes a person from the loop. The regulation says the opposite: baseline Part 107 puts a person in the loop by design, and the only way out is a waiver that somebody had to earn. When a vendor sells you autonomy, what they have usually built is a very good aircraft plus a paperwork position, and the paperwork position is the part that expires, has limits, and belongs to them.
And the aircraft is not quiet. It is required not to be. That is a strange property for a security asset, and it is worth deciding on purpose whether you want it, rather than discovering it after the dock is bolted to the lawn.
Scope note: this report describes federal aviation regulations as published and a single vendor's public statements about its own FAA authorisation. It is not legal advice, not an aviation compliance opinion, and not a security assessment of any property or product. Waiver scope, currency and limitations live in the waiver document itself, which was not obtained. Part 108 was a proposed rule as of the publication date above and its status may have changed; any correction will be made in place and dated. State privacy law, local ordinances, and property law governing aerial observation were outside the scope of this report and are frequently the binding constraint in practice. Confirm your own operation with a qualified aviation attorney and with the FAA directly.
U.S. Army combat veteran (Counter-IED / Electronic Warfare), peer-reviewed researcher in microwave spectroscopy, and founder & CEO of Shroombiosis. Consults on laboratory operations, AI, and supplement formulation.