Public transport without a fixed timetable: an autonomous on-demand shuttle service developed at TU Graz integrates vehicle and infrastructure sensors to enable safe, efficient and sustainable mobility.

The transport sector is still responsible for around a third of greenhouse gas emissions. Shifting from private transport – particularly vehicles with internal combustion engines – to electric public transport offers a way to reduce emissions. To this end, researchers at Graz University of Technology (TU Graz) have teamed up with Chinese partners to develop an on-demand mobility service available 24/7, which can bridge the gap between the convenience of private cars and the efficiency of public transport. At the heart of the project is a battery-powered, self-driving shuttle bus capable of carrying up to 15 people. Instead of running on fixed routes according to a timetable, it can be requested on an on-demand basis and operate flexibly within a wider route network. A test vehicle is already being used successfully on a test route in real-world traffic conditions in China, and demonstrator vehicles are being planned in Austria.

“Every car journey – especially one in a combustion-engine-powered vehicle – that you don’t have to make helps the environment and the climate,” says project manager Cornelia Lex from the Institute of Automotive Engineering at TU Graz. “Through our project, we aim to make it easier for people to give up their private journeys, as the on-demand shuttle offers them an equally flexible and, moreover, sustainable way of getting to their destination.”

Collaborative perception

Central to the safe autonomous control of vehicles is the so-called collaborative perception. The shuttle is equipped with lidar, radar and camera systems and also communicates via 5G with smart infrastructure, such as cameras positioned on challenging or blind spots along the road. This additional connectivity enables the vehicle to monitor its surroundings – including pedestrians and cyclists – in real time, even if they are not detectable by the shuttle’s on-board sensors. In the event of a network failure, the vehicle is equipped with an independent safety feature that brings it to a controlled stop.

Millions of simulation kilometres

The researchers carried out extensive testing of the vehicles’ safety concept in advance using high-fidelity simulations. To this end, real streetscapes, complete with buildings and infrastructure, were precisely recreated in order to virtually validate the algorithms for extreme and everyday situations. As the validation of autonomous vehicles requires the testing of countless scenarios and the simulation of millions of kilometres of driving, virtual testing is the only practical way to provide the necessary safety evidence. The simulations make it possible to test complex algorithms for safe route planning, object detection and the behaviour of other road users under controlled conditions before they are deployed in real-world traffic.

To ensure the vehicles can be deployed as flexibly as possible, the university research partners in China have developed a dynamic optimisation system for routes and charging processes, with the aim of minimising waiting times for passengers and maximising capacity utilisation. “It is precisely high-capacity utilisation that helps to reduce our carbon footprint,” says Cornelia Lex. “If several people travel to their destination in an electric shuttle rather than each driving alone in their own car, there are significant benefits in terms of air quality and road space.”

In addition to TU Graz, Tongji University, Wenzhou University and Seisys from China were involved in the project. The project was funded by the Austrian Research Promotion Agency FFG.