Changi: cameras on the stand, driverless tractors on the road
Changi has evidence of AI-assisted turnaround monitoring in Terminals 2 and 3 and remotely supervised driverless baggage tractors in live service. These are two distinct moves from trial technology into airport operations.
Singapore Changi Airport Terminal 1 seen from Jewel Changi Airport, June 2019. Photograph: Nigos / CC BY-SA 4.0 (cropped to fit). Shows the airport, not an AI implementation.
The picture at Singapore Changi
From a trial route to normal airside work
Autonomous vehicles are often announced long before they can work in routine airport traffic. Changi's baggage-tractor programme gives a more specific picture: years of testing, live operations on a defined terminal route and further fleet expansion planned.
The airport has also implemented Aircraft 360, using computer vision to monitor turnarounds at Terminals 2 and 3. A Changi executive reported a fall in last-minute gate changes, a meaningful operating measure if the comparison can be inspected.
Neither story removes people from the process: the tractors are remotely supervised and the value of a turnaround alert depends on staff acting in time.
Selected implementations
Two applications with operational evidence
01 / Deployed
The gate change that never has to happen
The operational problem
A late turnaround can force a last-minute change in aircraft gate assignment and disrupt passenger plans.
What the sources report
Changi Airport Group's chief executive said in July 2026 that Aircraft 360 computer vision was in use at Terminals 2 and 3 and that last-minute gate changes had fallen 60% in its first six months.
Why it matters
This is an outcome tied to a recognisable operational problem, although attribution and baseline matter.
What remains unproven
The reported result comes from the airport executive; no independent causal assessment or detailed comparator is provided.
Evidence supports a reported claim; it does not automatically establish a measured operating benefit.
02 / Deployed · limited scope
The baggage tractor enters regular service
The operational problem
Repetitive towing between terminals consumes staff time and needs safe movement through a shared airside environment.
What the sources report
Changi and aviation-industry sources reported driverless baggage tractors operating on the Terminal 1–Terminal 4 route after more than 5,000 trial trips, with remote supervision and further vehicles planned.
Why it matters
The distinction is the documented step from supervised tests to service on a particular route.
What remains unproven
A full cost or intervention-rate comparison with conventional tractors has not been published.
Aurrigo's separate four-vehicle Auto-DollyTug programme remains a pilot for underwing baggage work. It should not be counted as another already scaled driverless service.
Aurrigo Auto-DollyTug pilot at Changi, four vehicles · Pilot
Changi Airport Group and Aurrigo agreed to bring in four autonomous Auto-DollyTugs to pilot underwing operations, Aurrigo announced on 14 May 2024.
The last two of the four vehicles were delivered in January 2025, Aurrigo said in its results statement of 19 May 2025, and all-weather testing under fleet operations was then under way.
Evidence supports a reported claim; it does not automatically establish a measured operating benefit.
Here the deployment stage really matters
Changi's evidence separates three stages unusually well: a turn-monitoring system in use, a defined live autonomous tractor route and a separate underwing vehicle trial.
The next valuable results would be attributable gate reliability, supervisor interventions and cost per baggage movement.
Live data from the Aerostack register. Status reflects published evidence only. See Our approach for how records are graded.
Aerostack register · 3 tracked deployments
Aerostack tracks AI deployments across airports and airlines, with the published evidence behind each record. Photography is real aviation imagery. Record thumbnails may show the named airport or airline; other photographs do not.
