What the evidence on Drone as First Responder actually shows

Claims about Drone as First Responder are easy to find. Published evidence is harder. Most of the numbers circulating in the market come from vendor pages and conference slides, and a procurement committee will treat them accordingly.

That picture changed a little in September, when a study of one of the largest programmes in service appeared in a peer-reviewed journal. This post sets out what has actually been measured, where the figures come from, and which claims still have nothing behind them.

The strongest dataset so far

In September 2026, Policing: A Journal of Policy and Practice, published by Oxford University Press, carried a study of the Dubai Police Drone Box system, written by researchers at the Dubai Police Academy together with the force's General Department of Operations.

The study covers the whole of 2023 and every drone-assisted response in that period: 1,502 incidents, of which 1,496 were classified by severity. That matters methodologically, because it is the complete population and not a sample, which removes the question of how cases were selected.

The response times are consistent across incident types. Mean arrival was 1 minute 25 seconds for incidents triaged as very critical, 1 minute 28 for critical, and 1 minute 26 for normal. Medians sit within a second or two of the means, standard deviations stay under 11 seconds, and the slowest single response recorded was 2 minutes 6 seconds.

The measurement starts at the trigger, which is the emergency call, a CCTV detection or a system alert, and ends at confirmed arrival on scene. Timestamps are logged automatically by the platform, so the figures do not depend on anyone remembering to write a time down.

The consistency is worth more than the average. A system landing between 1:00 and 2:06 across fifteen hundred incidents is behaving predictably, and predictability is what an operations manager plans around.

The operation runs beyond visual line of sight from a central control room, around the clock, under a specific operations risk assessment submitted for each location and licensed by the Dubai Civil Aviation Authority.

The scale is worth knowing, because it puts the incident count in proportion. In the period studied the programme ran from eight Drone Boxes plus one mobile unit, with planned expansion towards 800 square kilometres of coverage. So 1,502 responses in a year came from a network of that size, which is a more useful benchmark for anyone sizing their own programme than the response time on its own.

What American programmes publish about themselves

Two US forces publish running figures, and the Dubai study cites both.

Brookhaven Police in Georgia reported responding to 1,700 calls for service in 2024, with the drone first on scene for 72 per cent of them and an average response time under two minutes. Chula Vista Police in California, the longest-running programme of its kind, publishes a live dashboard: between 1 January and 28 October 2025 the force responded to 2,317 calls, the drone arrived first at 943 of them, average response time was 2.4 minutes, and the programme avoided dispatching 109 patrol units.

Both forces publish their own operational data and keep it updated, so anyone can check the current figures directly.

The medical parallel

The closest thing to a European controlled comparison comes from emergency medicine. A Swedish study published in Resuscitation in 2024 compared drones dispatched to time-critical incidents against ground EMS. Across 52 cases, and despite unfavourable weather and closed airspace in some of them, median drone response was 3 minutes 49 seconds against 5 minutes 51 for the ambulance, and the drone arrived first in 47 of the 52.

Slower than the policing figures, in a different operating model, and peer-reviewed. For anyone arguing a case to a sceptical board, a published European study showing the same directional result is worth having alongside the policing numbers.

What has not been established

Four things are missing from the literature.

Cost savings are not quantified. The Dubai authors name this explicitly as missing: nobody has yet calculated the financial comparison between sending a drone and sending a patrol officer, including the time lost when an officer attends an incident that did not need them. Every euro figure you will see on this subject is currently a vendor's model.

The effect on crime is unmeasured. Changes in crime types and rates before and after a drone programme would be valuable, and the study lists it as future work.

Nothing is longitudinal. All the strong data is a single year or less, at a single site. Trends over time, and whether the early performance holds as a programme scales, remain open questions.

Call-clearance rates vary wildly. The share of calls resolved without a ground unit is reported anywhere from about a fifth to more than half, depending on who is publishing and which calls are counted. Chula Vista reports 109 patrol units not dispatched over its ten-month period, and that figure counts units and not calls, so it does not map onto those percentages at all.

What to do with this

Response time is the claim that holds up. Sub-two-minute arrival is documented across three programmes on two continents, one of them in a peer-reviewed journal with a full year of data. That is the number to quote, with the source attached.

Everything financial is still unmeasured. No cost comparison between a drone response and a patrol response has been published anywhere, so any figure on this subject is a model and not a finding.

The gap is European data. Every policing dataset here comes from Dubai or the United States. No European force has published comparable response-time figures, which means operators here are currently arguing from other people's jurisdictions.

A last point on Dubai. The published study documents the performance of the Drone Box system as operated by Dubai Police, and it does not assess the software behind any particular deployment. Dubai Police is an AirHub customer and we are proud of the work, and the study stands on its own as evidence about what a mature dock-based programme can achieve, which is how it should be read.

If you want to talk through what this means for a programme in your own jurisdiction, our public safety page sets out how the platform supports each stage, and our regulatory consultancy team works with operators on the authorisation route.

*Picture: Dubai Police Drone Box (Source: Dubai Police UASC 2025).