Search across all content
An on-demand transport pilot that used AI to route vehicles in real time to passenger requests, tested in a Hungarian city as an alternative to fixed bus routes.
Public transport in smaller cities runs on fixed routes at fixed times. That makes sense when passenger demand is predictable and spread evenly across a network. But in smaller cities, demand is often uneven. Some routes carry passengers at certain times of day and run nearly empty at others. Some parts of the city are underserved. The bus comes whether or not there are passengers waiting, resulting in empty runs and unnecessary emissions.
Gödöllő, a city of under 40,000 people near Budapest in Hungary, had faced this problem for years. Its public bus network consisted of five buses running 12 fixed routes. Average wait times for residents were around 60 minutes. As one resident later described it: "Without a car or driving license, it used to be difficult to get from one place to another."
This is not unique to Gödöllő. Smaller cities across Europe and beyond face the same challenge. The population is not large enough to sustain frequent, comprehensive services, but residents still need reliable access to work, healthcare, education, and shops. The options available to local authorities have traditionally been to add more vehicles, extend existing routes, or accept that some areas will remain underserved. Each involves trade-offs between cost, efficiency, and the level of service residents receive. For many residents, the result was simple: a public transport system that existed in theory, but was not a realistic way to get around.
In 2025, Gödöllő became the site of a pilot that approached the problem differently. Rather than running buses on fixed routes, the city trialled Shucle, an AI-powered transport platform developed by Hyundai Motor Company. Shucle responds to ride requests from passengers in real time. The AI calculates the most efficient route to pick up and drop off multiple passengers, adjusting continuously as new requests come in. The concept is known as demand-responsive transport. Passengers request a ride through an app, and the system works out how to serve them alongside other passengers heading in a similar direction at a similar time. There are no fixed stops or timetables. The vehicle goes where it is needed, when it is needed.
For the Gödöllő pilot, two Shucle vehicles were integrated into the city's existing bus network for 11 weeks from August to October 2025. Hyundai collaborated with the municipality and local bus operators to integrate the on-demand service into Gödöllő's existing network of five buses and 12 fixed routes. The pilot was part of the Economic Innovation Partnership Program, a collaboration between the Korean and Hungarian governments that provides policy and technical support to partner countries.
The platform had been running in several South Korean cities and towns since 2021, but Gödöllő was its first deployment outside the country.
1. Average wait times fell from 60 minutes to 6 minutes
Before the pilot, residents using Gödöllő's public bus network waited an average of 60 minutes for a bus. During the 11-week Shucle pilot, that dropped to 6 minutes, a reduction of more than 90%. A total of 2,950 residents used the service, making 3,138 ride requests. For a city where public transport had not been a practical option for many daily journeys, the change was significant. As one resident described it: "Without a car or driving licence, it used to be difficult to get from one place to another, but thanks to Shucle, we can now move freely around town."
2. Two on-demand vehicles covered more of the city than five fixed-route buses
Gödöllő's existing bus network ran five vehicles across 12 fixed routes. Those routes were set regardless of where passengers needed to go on any given day, which meant some parts of the city were well served, and others were not. The two Shucle vehicles took a different approach: instead of following a predetermined route, they adjusted in real time based on where ride requests were coming from. The result was that two vehicles reached parts of the city that five buses on fixed routes might not.
3. Fewer empty runs, lower emissions per journey
A persistent inefficiency in fixed-route bus systems is that vehicles run whether or not anyone is on board. During off-peak hours or on less popular routes, buses can run nearly empty for much of the day. Because Shucle only dispatches vehicles in response to actual ride requests, it reduces the number of journeys where a vehicle is running without passengers. Fewer empty runs mean less fuel consumed and lower emissions. The pilot was conducted under a programme explicitly linked to Hungary's carbon neutrality goals.
In smaller cities, how vehicles are deployed may matter more than how many there are. The pilot suggests that matching routes to real-time demand can produce a more responsive service than adding vehicles to fixed schedules. For transport planners in cities with lower or more dispersed demand, demand-responsive models are worth considering alongside conventional route expansion.
The technology needed local adaptation to work. The platform was not imported unchanged from South Korea. Hyundai worked with the municipality and local operators to design the service around Gödöllő's specific transport patterns. For governments considering adopting technology developed elsewhere, the pilot illustrates that collaboration with local partners during setup is part of making it work.
The evidence base is still early. Whether the results hold in larger cities, over longer periods, or under different conditions has not yet been established. Hyundai has stated that the pilot will inform further deployment elsewhere.
Launch year: 2025
This case study was written with assistance from artificial intelligence.





Connect with 500,000+ public servants solving your hardest challenges.





Connect with 500,000+ public servants solving your hardest challenges.
Help public servants worldwide learn from your work, what worked, what flopped and what you'd do differently
Share your project
Log in or sign up to continue the conversation