This page will not run the usual argument, because Robin Radar’s published material refutes it. The familiar counter-UAS line — that radar was built for large fast aircraft and misses small, slow, hovering drones — does not describe IRIS. Robin Radar states that IRIS uses “micro-doppler classification and DNN technology” to distinguish blades and rotating parts, and that “IRIS isn’t phased by birds and detects fixed-wing and hovering drones”. It is a purpose-built drone radar and the small-drone case is what it was designed for. Any honest comparison has to start there.
So the difference is not blindness. It is emission, and everything downstream of emission. A radar detects by transmitting energy and listening for the echo: IRIS is published as an X-band, FMCW radar with a “2x 12 w” power output, a rotating head at “30 rpm / 1.0s”, and a 5 km instrumented range covering 78 km². GOTEAM detects by receiving heat the aircraft is already radiating, and transmits nothing at all. GOTEAM’s published position is that it therefore “does not require a spectrum license in any jurisdiction (because it does not emit)”, cannot be jammed or geolocated by the target, and adds nothing to a site’s RF noise floor.
The second difference is what comes out. A radar returns a track: range, bearing, elevation, velocity, a classification. GOTEAM returns a track and the frames that produced it — an annotated clip and zoom-crop thumbnails an operator can look at before acting, and re-run later against a newer model from the node’s raw frame archive. Radar measures better. Imagery shows you.
IRIS figures quoted here are from Robin Radar’s own published specification. Rows that are not quoted are written at the level of the technique — active radar versus passive thermal imaging — which is true by physics and needs no vendor citation. See the note at the foot of the page.
Most of this table is one property restated: GOTEAM adds no energy to the environment and produces a picture. Those two facts drive the licensing position, the covertness argument, the absence of RF interference and the reviewability of every detection after the fact.
| Capability | GOTEAM | Active radar |
|---|---|---|
| Emits nothing — the sensor is undetectable and cannot be direction-found by the target | ✓passive EO/IR; no radar pulses, no RF | ✗IRIS is published with a “2x 12 w” X-band output; transmitting is how radar works |
| No spectrum licence required, in any jurisdiction | ✓GOTEAM’s stated position, and a direct consequence of not emitting | ✗an X-band transmitter generally requires spectrum authorisation; check locally |
| Adds nothing to the local RF noise floor | ✓matters at airports, ports and sites where the spectrum is already contested | ✗ |
| Every detection has reviewable imagery attached | ✓annotated full-length clip and zoom-crop thumbnails per track | ✗a radar returns a track, not a picture; imagery requires a separate cued camera |
| Detections re-runnable against a newer AI model, from archived raw frames | ✓one-click reprocess from the node-side raw frame archive; new model published fleet-wide and hot-swapped in seconds | — |
| No moving parts in the detection path | ✓fixed stare — no gimbal motors, slip rings or cryocooler, so no mechanical wear items | ✗IRIS is published as a rotating head at “30 rpm” |
| Detection engine licensable as software onto a third party’s thermal core and compute | ✓camera-agnostic; Jetson or x86 with Intel Arc | — |
| Fits a covert or politically sensitive deployment where an emitter is unacceptable | ✓no emission to detect, license, or explain | ✗ |
Both systems detect small drones, separate them from birds, track multiple targets and hand a threat picture to an operator. Ticking both columns would be true and worthless. The interesting content is how, because the how decides where each one fails.
| GOTEAM does it by… | Robin Radar does it by… | |
|---|---|---|
| Detecting a small drone | Thermal contrast. Sixteen-bit uncooled LWIR frames, morphological sky and cloud suppression, then a neural detector trained on tens of thousands of hand-verified thermal drone frames that proposes targets a handful of pixels across. | Radar return. An X-band FMCW waveform with a one-second update over 360° azimuth and 60° elevation, out to a published 5 km instrumented range. |
| Separating drones from birds | Shape, then behaviour. Neural classification on the imagery, flight-behaviour cues that down-weight birds, and a tracker that confirms a target only once it persists and moves consistently. | Micro-Doppler. “Classification with micro-doppler and deep neural network (DNN)” — reading the modulation that rotating blades impose on the return, which works with no pixels on target at all. |
| Handling many targets at once | A fixed stare. The whole field of view is imaged and searched on every frame, so the AI resolves every target simultaneously — one drone or twenty, at the same latency. | A full-volume scan. 360° coverage with “speedy update rate of one second”, tracking everything in a 78 km² footprint from a single 29 kg unit. |
| Giving the operator range and bearing | Estimated from the image. Per-target range, bearing and elevation derived from box size and camera field of view, with a positional-uncertainty ellipse on the map — an inference, and presented as one. | Measured directly. Range comes out of the waveform, not out of an assumption about how large the target is. This is the cleaner quantity and there is no honest way to argue otherwise. |
| Feeding command-and-control | SAPIENT / BSI Flex 335 v2, as both server and client, plus JSON over a local interface — detections to higher command, partner sensors on the same map. | An integration-first radar. Robin Radar: “IRIS seamlessly integrates with command-and-control (C2) platforms and wider C-UAS ecosystems”. |
Where GOTEAM is stronger. It is silent, and the silence is the product. A site that cannot get a spectrum licence, cannot accept an emitter on a contested RF floor, or cannot afford to tell an adversary that it is being watched, has a very short list of options and a rotating X-band transmitter is not on it. GOTEAM also returns the one thing a radar structurally cannot: the frames. When a detection has to be justified — to a commander, a regulator, an inquiry — an operator who can watch the clip is in a different position from one holding a track file.
Where Robin Radar is stronger. Coverage, weather and measurement. Five kilometres of instrumented range over 78 km² from a single 29 kg head, switchable to twelve, is a volume a passive thermal node does not address; micro-Doppler classifies at ranges where an imager has nothing to look at; and none of it cares whether the air is clear. Radar also gives range as a measured quantity rather than an inference from apparent size. Against a large perimeter, or on a day with cloud on the deck, that is decisive.
Use both — this is the textbook cue-and-confirm pair. A volume-search radar is at its best finding something far out and telling you where to look; a passive imager is at its best confirming what it is, silently, with footage attached. GOTEAM’s published position is that it is a layer that fuses with radar, RF and acoustic sensors through an open API and SAPIENT (BSI Flex 335 v2), and against a radar of this class that is not a hedge — it is the correct architecture.
IRIS figures and all quoted phrases on this page are taken from Robin Radar Systems’ own published material as of September 2026 — robinradar.com/products/iris-radar and its radar products overview. Last reviewed: October 2026. ◐ = partial — present but narrower than the other column. — = not documented publicly: a capability Robin Radar’s published material does not address at this level of detail, not a claim that it is absent. GOTEAM marks reflect capability published on this site and in /llms.txt; specifications beyond that are shared under NDA after vetting. Rows written at the level of the technique — that a radar transmits and a thermal receiver does not, that an imager needs an optical path — are properties of the physics and apply to every product of that class, GOTEAM’s included. Robin Radar, IRIS and MAX are trademarks of Robin Radar Systems B.V.; this is an independent capability comparison, not an endorsement, and every product here evolves — verify specifics against current documentation.
GOTEAM nodes report over SAPIENT (BSI Flex 335 v2) as both server and client, so partner sensors and GOTEAM tracks share one map.