The questions that actually separate SAR drones

Pick up any drone buyer's guide and you'll find the same checklist — flight time, camera resolution, range, price. That list is fine for hobbyists. It's useless for search and rescue. SAR missions don't fail because a drone has 28 minutes of endurance instead of 35. They fail because the operator couldn't see through smoke, couldn't fly in 12 m/s gusts, couldn't legally deploy at a federal site, or couldn't connect the drone to the autonomy software that actually finds the missing person.

After building Overwatch on four different industrial platforms and watching customers evaluate a dozen more, we've converged on a short list of questions that genuinely predict whether a drone will earn its keep in a SAR role. Everything else is specification-sheet theatre.

The first question is thermal. Not "does it have a thermal option" but "is the thermal camera genuinely useable for detecting a human body at realistic search altitudes". A 320×256 microbolometer radiometric sensor on a gimbal that can point 90° down is the minimum viable SAR payload. Lower resolution turns a person-sized heat signature into a single pixel the operator will miss. Higher resolution (640×512 or FLIR Boson+ class) roughly doubles the reliable detection altitude, which in turn doubles the area you can sweep per battery. Thermal isn't a nice-to-have on a SAR drone. It's the whole point of flying at night, in fog, or over dense canopy.

The second question is weather rating. Missing people don't wait for clear skies. A drone with an IP43 rating or lower is a fair-weather asset — useful for planned patrols, useless when the coastguard gets a call at 02:00 in driving rain. IP54 is the threshold where SAR teams start trusting a drone to launch without checking the forecast. IP55 or above lets you operate in the conditions that actually generate SAR call-outs in the first place.

The third is open SDK access. This is the question almost every marketing page obscures. A drone with a locked flight-control stack can only ever do what its manufacturer's app lets it do. If you want autonomous grid search, relay handoff between multiple drones, onboard detection running on a companion computer, or integration with your dispatch system, you need ground-to-drone command access — MAVLink over UDP or serial, or an SDK that lets you inject mission waypoints, change parameters mid-flight, and pull telemetry at reasonable rates. The closed ecosystems (we'll name them shortly) make a Faustian bargain: excellent out-of-the-box flight behaviour in exchange for a permanent ceiling on what third-party software can add. For a SAR programme planning to evolve over five years, that ceiling is the deciding factor.

The fourth is procurement compliance. US federal agencies can't operate DJI drones — the Countering CCP Drones Act and DoD Section 848 effectively bar them from federal SAR. Blue UAS certification (the DoD-approved list) is increasingly the procurement floor for US state and municipal agencies too, and European defence procurement is trending in the same direction. If you're a coastguard, police force, or military SAR unit in the US, EU, or UK, the NDAA / Blue UAS / EU-sovereignty question is not optional. It's the first filter, applied before any technical evaluation.

The fifth is endurance under load. Manufacturers quote endurance in clean, windless, unpayloaded conditions. A SAR drone with a thermal gimbal, RTK GPS, a companion computer, and 8 m/s of headwind will deliver 55-70% of the glossy-brochure figure. Ask for the honest number with the payload you actually intend to fly. Then cross-reference that against your planned patrol duration and decide whether you need a single drone, a two-drone relay, or a larger fleet. Large AOIs almost always need multiple drones regardless of which platform you pick.

Parrot ANAFI USA and ANAFI UKR

Parrot sits at the approachable end of the industrial SAR market. The ANAFI USA is Blue UAS listed, which clears it for US federal operation and makes it one of the few genuinely turnkey compliant options under €10k. The UKR variant is the EU-facing partnership build with the same core hardware — Snapdragon 845 compute, 32× hybrid zoom, FLIR Boson thermal, roughly 32 minutes endurance, IP53 weather rating.

The strength is compactness and compliance. The weakness is the SDK model. Parrot's AirSDK is a mission-uploader architecture — you package a flight plan, upload it over WiFi, and the drone flies autonomously offline. That works brilliantly for single-drone grid search missions and for customers who want an offline-first air-gapped workflow. It does not work for fleet patrol, because the drone is sealed once it launches. No mid-flight parameter tuning, no ground-to-drone abort from a central operator, no relay handoff. This is why Overwatch runs single-drone SAR mode on ANAFI UKR but reserves the fleet-patrol product for MAVLink platforms — the physics of the Parrot SDK makes relay orchestration impossible.

If your mission profile is a single operator with a single drone, handheld controller, and a reasonable AOI, the ANAFI USA / UKR is a defensible choice. If you're building a coastguard patrol that runs 24/7 with drones rotating through a charging base, it isn't.

Freefly Astro Max

The Astro Max is where the US industrial-drone market has quietly landed for agencies that need a Blue UAS-certified platform with a genuinely open flight stack. It's built on Auterion's Skynode, which is PX4 underneath, which means full MAVLink access out of the box. You get mid-flight telemetry, mid-flight commands, parameter tuning, mission re-upload — the full ground-station toolkit.

The payload story is the Smart Dovetail mount, which is an open mechanical and electrical interface. That means thermal payloads from multiple vendors (FLIR, Teledyne, third parties), your own custom payloads if you need them, and no vendor-lock on the imaging side. Endurance with the standard battery is around 37 minutes unloaded, roughly 28 minutes with a full thermal-and-visible gimbal. IP rating is IP43, which is the Astro's weakest spec — acceptable for planned missions, marginal for all-weather response.

For US federal SAR and any agency that needs Blue UAS plus open autonomy, the Astro Max is the reference choice. It's the platform we recommend first when customers ask "what should we buy".

Skydio X10

We don't support the X10 in Overwatch. In fairness to Skydio, they make genuinely impressive hardware — the onboard obstacle avoidance is best-in-class, the thermal option (X10T with the FLIR Boson+ 640) is excellent, and the NightSense low-light camera is a real SAR tool in its own right. The X10 is Blue UAS listed, IP55 rated, and endurance is competitive at around 35 minutes.

The problem is the SDK. Skydio's autonomy is their competitive moat, and they've chosen to keep the flight stack closed. You can integrate at the mission-planning level through their cloud API, but you cannot run a third-party autonomy stack on the drone, you cannot connect a companion computer to the flight controller, and you cannot do the kind of deep integration that fleet patrol requires. If Skydio's own flight behaviour matches your mission exactly, the X10 is superb. If you want to build on top of it — custom detection models, relay handoff, integration with your CAD or dispatch system — you'll hit the ceiling quickly.

Our honest position: if you're a single-drone SAR operator and Skydio's out-of-the-box autonomy does what you need, buy the X10. If your five-year plan involves orchestrated fleets or integration with external systems, buy something with MAVLink.

DJI Matrice 30T

The M30T is the elephant in the room. It's the most widely deployed SAR drone in the world, its thermal-plus-visible-plus-laser-rangefinder gimbal is genuinely excellent, endurance is 41 minutes, IP55 rated, and the cost-per-capability ratio is unmatched. For a European fire brigade or a non-US coastguard with no procurement-compliance constraints, the M30T is an easy recommendation on technical merit alone.

The problem is political and regulatory. DJI is effectively barred from US federal use. Several US states have followed. The UK's CAA hasn't restricted DJI, but UK police and military procurement increasingly prefer Blue UAS platforms for interoperability with US allies. EU defence procurement is drifting the same way, driven partly by the war in Ukraine and partly by long-running concerns about data sovereignty. DJI's SDK is also closed — the Payload SDK gives you camera and gimbal access but not flight control, and the Onboard SDK has been progressively restricted on newer models.

If you're a civilian agency in a jurisdiction with no DJI restrictions and no plans to integrate with third-party autonomy, the M30T is a perfectly rational choice. We don't support it because the customers we serve — coastguards, police, military SAR — almost universally need a compliant, open alternative.

Sky-Hopper

Sky-Hopper is the dark-horse recommendation. Israeli-made, purpose-built for SAR and tactical surveillance rather than adapted from a commercial mapping drone. The standard payload is a dual thermal-and-visible gimbal that is genuinely SAR-optimised — high-resolution thermal, good low-light visible, stabilisation tuned for stationary-target identification rather than cinematic smoothness. Endurance is competitive, IP rating is solid, and the flight stack is MAVLink-accessible.

The procurement story is nuanced. Sky-Hopper isn't Blue UAS, so it's a harder sell for US federal use, but it's well-received in European and Middle Eastern markets and for agencies that want a combat-proven SAR-specific platform. If your procurement framework permits Israeli hardware, Sky-Hopper deserves a serious look — it's the only platform on this list designed from day one for exactly the mission Overwatch orchestrates.

Quantum Systems

Quantum Systems is the European sovereignty answer. German-built (increasingly important as EU procurement tilts toward domestic suppliers), MAVLink-compatible, and with a unique platform mix — the Trinity Pro VTOL fixed-wing and the Vector VTOL both offer endurance far beyond multirotors, which matters enormously for large-AOI coastal patrol and wilderness SAR. A Trinity on a two-hour flight covers ground that would take a multirotor fleet of eight drones to match.

The trade-off is that VTOL fixed-wings don't hover, so close-in inspection of a detected target requires either a secondary multirotor or acceptance that you're doing detection-only from altitude. For pure search, the endurance advantage is enormous. For search-and-investigate, a mixed fleet of Trinity plus multirotor is the right pattern — and this is exactly the kind of heterogeneous fleet that a ground-station layer has to orchestrate.

How Overwatch fits in

Here's the uncomfortable truth about the platform comparison above: the hardware differences matter less than the software differences. A Freefly Astro Max flying the manufacturer's default flight software and a Sky-Hopper flying the manufacturer's default flight software will both conduct a perfectly competent manually-piloted search. Neither will, out of the box, run a boustrophedon grid search over an operator-drawn AOI with automatic sector partitioning, relay handoff between multiple drones, onboard human detection feeding notifications to your dispatch channel, and a weather-aware safety gate that triggers an automatic return when wind crosses 10 m/s.

That autonomy layer is what Overwatch is. We run it on whichever of these platforms you've chosen — ANAFI UKR for the AirSDK path, Astro Max or Sky-Hopper or Quantum Systems for the MAVLink path — and we deliver the same mission experience across all of them. Single-drone SAR missions look the same to the operator regardless of airframe. Fleet patrol with automatic sector partitioning and relay handoff looks the same across any supported MAVLink fleet. The onboard detection pipeline runs on the companion computer, not the flight controller, which means it's platform-agnostic. The Overwatch Ground Station connects to whatever MAVLink source is feeding it.

That platform-agnosticism is deliberate. We don't want to be the software for one drone. We want to be the autonomy layer for whatever drone your procurement, your jurisdiction, and your mission profile tell you to buy. If that's a Freefly Astro Max because you need Blue UAS, fine. If it's a Sky-Hopper because you need SAR-specific thermal, fine. If it's a Quantum Systems VTOL because you need two hours of endurance over a coastline, fine. If it's an ANAFI UKR because you need a compact offline-first single-drone asset, fine. The hard work of search pattern generation, fleet orchestration, detection, alerting, and safety gating is the same work regardless of which airframe is carrying it.

Pick the drone that matches your compliance requirements, your weather envelope, and your payload needs. Then put an autonomy layer on top that turns it from a camera on a stick into a system that actually finds missing people. That's the decomposition that works. The alternative — picking a drone for its built-in autonomy and hoping the ceiling is high enough — is how agencies end up replacing their entire fleet every three years because the software couldn't keep up with the mission.

Ready to evaluate Overwatch? See pricing and platform options or book a 30-minute consultation.