We Took A Ride In A University AV: Here’s What We Learned

(From left to right: Gamal Elghazaly; Research Scientist, Raphaël Frank, Senior Research Scientist and Mehdi Testouriour; Research and Development Specialist at SnT © Stephanie Jabardo / Silicon Luxembourg)
(From left to right: Gamal Elghazaly; Research Scientist, Raphaël Frank, Senior Research Scientist and Mehdi Testouriour; Research and Development Specialist at SnT © Stephanie Jabardo / Silicon Luxembourg)

A ride in “Junior”, the University of Luxembourg’s autonomous car, shows both the promise and the challenges of the technology set to shape Luxembourg’s future mobility landscape.

At the car park of the University of Luxembourg, I and two colleagues pile into “Junior”, the university’s autonomous vehicle, built at an estimated cost of €50,000. Aside from the lidar on its roof, which scans the surroundings to create a 3D model of the environment, Junior looks like a normal, if slightly dirty, electric Kia Soul. “It cannot be cleaned at a regular car wash because of the technology on it,” explains Raphaël Frank. “We need to motivate our students to clean it by hand.”

Prof. Frank is a Senior Research Scientist at the Interdisciplinary Centre for Security, Reliability, and Trust (SnT) and leads UBIX, SnT’s research group dedicated to distributed artificial intelligence across fields including automated vehicles.

Junior’s interior is busy. In addition to the standard dashboard, there is a computer screen; a camera hangs from the windshield, one of seven the vehicle uses. Mehdi Testouri, research and development specialist at the SnT sits behind the steering wheel, while Frank takes the passenger seat, equipped with a keyboard that includes an emergency brake. Between them sits another red emergency-stop button.

“Just so you know, this is an experimental vehicle, something can go wrong. We need to do all these procedures because our system is not perfect,” Frank says.

Before we set out, we hit a small snag. Our bags and coats, kept in the cabin because the boot houses the AI and onboard computer, appear to have dislodged a cable.

Once the system is running, Junior cruises along Boulevard Pierre Frieden at a sedate 30 kph, the steering wheel turning as if by ghost hands. Its movement is jumpy, like that of a first-time driver, as its cameras and sensors process signs, traffic lights, lanes, vehicles and pedestrians. The vehicle relies on high-definition maps developed in-house. But when a bus is parked along the mapped path, the system becomes confused and Mehdi takes over manually.

“I’m used to it now, but it was a bit scary at the beginning,” Mehdi says of driving an automated vehicle. His hands frequently hover near the steering wheel, though Junior continues to drive itself. He keeps them close as a precaution. By now a steady stream of vehicles has formed behind us. We turn onto Boulevard Konrad Adenauer, where cars overtake us and Junior slows with each pass. Fortunately, research scientist Gamal Elghazaly follows in a support car, preventing any dangerous build-up behind.

“[…] skills can be applied to different sectors, not necessarily only autonomous cars.”

Prof. Raphaël Frank, Senior Research Scientist at the Interdisciplinary Centre for Security, Reliability, and Trust (SnT)

Junior and the test loop are not new. What is new is the team’s teleoperated system, a remote simulator controlled by a human, designed to take over an autonomous vehicle if the onboard system fails. Before we got in the car, the team demonstrated the simulator, which uses the University’s internet and the 5G network outside. On a screen, we watched a live view from Junior as it was manoeuvred remotely down a path in the car park. Challenges remain. Frank lamented the video latency, but also the fact that the simulator driver needs to feel they are in the vehicle to take full responsibility. “The idea is to use a headset with a 360° camera to look around,” he says. “And the platform has a motor that could take vibrations from the car and connect it to the seat.”

The research project aims to educate the students, who help implement new features. As Luxembourg aims for a nationwide roll-out of level 4 high automation systems with no driver intervention by 2028, these are skills and knowledge the country needs. Two alumni are already working in the sector: one is working as an engineer with Ohmio, the New Zealand company operating an autonomous shuttle in Belval. Another former student whose thesis focused on mobile robots is today working for autonomous solar panel cleaning firm Solarcleano. “These skills can be applied to different sectors, not necessarily only autonomous cars,” says Prof. Frank.

It is not clear how the deployment will look in Luxembourg. In the US and China, robotaxis are widely deployed. Luxembourg says it wants autonomous vehicles for robotaxis, last-mile automated shuttles, automated valet parking, automation in logistics for freight and autonomous driving on motorways. Deployment will require high-definition maps, which are regularly updated. Prof Frank reckoned that we will see third party companies creating these in the coming years.

Another barrier to deployment is adoption and public trust. China has shown high adoption levels thanks to strong consumer demand, a regulatory push and an ecosystem that encourages innovation within China. A strong legal framework will play a key role here. Currently, France has the most advanced legal framework in Europe and the G7 for deploying automated vehicles and road mobility services. Luxembourg is in the process of developing new legislation on how to apply for authorisation to drive autonomous vehicles. Prof Frank says multiple laws will follow for each use-case scenario, but a first draft law is expected before the end of 2025.

Total
0
Shares
Related Posts
Total
0
Share