For the first time, an autonomous family car has weaved through traffic in Luxembourg. The demonstration, which took place in Kirchberg, is the culmination of five years of research by the 360Lab team at the University of Luxembourg. Technical research and legal developments will enable the transition from the current level of conditional autonomy to full autonomy in the future.
Artificial intelligence for a safer and more sustainable ecosystem
At the start of November 2022, the Interdisciplinary Centre for Security, Reliability and Trust (SnT) of the University of Luxembourg demonstrated its autonomous car in live traffic on the Kirchberg plateau. This vehicle is the test platform for navigation technologies and high definition (HD) maps which are currently under research at the 360Lab of the SnT.
The car, a specially adapted electric Kia, was driven on a 3km circuit. It has state-of-the art technology, including a sensor to map the road in real time, an on-board computer with trained artificial intelligence and an HD map. In fact, HD maps are the foundation of autonomous navigation: they are exponentially more detailed than the standard definition (SD) maps used by traditional navigation systems in cars and smartphones currently on the market.
Three questions for... Prof. Dr. Raphaël Frank, director of the 360Lab
How did the autonomous driving test go? Did new avenues of research open up as a result of the demonstration?
The purpose of the event was to show our partners and the general public the various 360Lab projects focusing on intelligent mobility and, in this particular case, autonomous driving. As such, it was mainly an event to showcase the progress of our research as opposed to a purely scientific test. The test went smoothly and there were no major problems.
And yes, new avenues of research have opened up. This test mainly demonstrated an engineering project, but there are many small pockets of research linked to the autonomous car that 360Lab PhD students are already working on.
One of these areas of research is the development of highly detailed mapping. It is like a version of Google Maps for autonomous vehicles, which provides extremely precise data on the environment. This technology will considerably reduce the costs of this type of vehicle because it will no longer be necessary to have sensors, which are often very expensive pieces of equipment. The aim is to have a lot of data about the environment already available in the car or that can be downloaded via a mobile network, such as 5G. The purpose of this research is to ensure that the mapping is always up to date as well as to make sure that the information on the map corresponds to the reality on the ground. These different approaches are being studied at 360Lab.
How can autonomous driving deliver a safer and more sustainable mobility ecosystem?
We work on the principle that robots, and therefore the autonomous car, will be much safer than a vehicle driven by a human.
Looking further into the future, when cars are fully autonomous, we will no longer need manual driving (or significantly less). As such, the number of cars on our roads may be reduced considerably. This also means that we will not necessarily have to buy or own a car. Instead, it would be a mobility-on-demand service: there will be vehicles located everywhere and when you need to travel from point A to point B, you would be able to call upon a whole range of mobility options. It would follow a similar model to Uber - we will have mobile apps to book a ride and these cars will pick us up and drive us autonomously to the destination of our choice. By using this approach, the aim is to use cars as much as possible and thus reduce the total number of cars on the roads. As a matter of fact, the majority of cars are just used for one hour a day and then sit idle for the rest of the day. Using this approach, we would be much more effective by embracing the idea of sustainability and safety based on robots.
Can autonomous mobility technologies be used in other areas of research? Do they have other applications in everyday life?
There is a huge potential for these applications. Much of the technology behind the autonomous car is perception-sensing technology: try to recognize what is around the car. I have already mentioned digital mapping, but this is only used for static objects, which do not move. But you also need sensors to detect moving objects around the car. For this, there are vision-based systems: for example, cameras which detect other cars and try to monitor them. These sensing technologies are already used in many artificial intelligence applications for image analysis. For example, if you enter a car park, there are now often cameras that will read the number plate and, when exiting the car park, the camera will automatically detect the number plate and open the barrier. And if you've already paid, you no longer need a ticket. That is one example, but there are thousands of others. Perception-sensing technology is already used in many fields.
All areas related to robotics, such as autonomous drones and the space industry already use very similar technologies.
Classification of autonomous driving
SAE International is an international trade organisation that develops standards for the transportation industries, including aerospace, automotive and commercial vehicles. For the autonomous driving of vehicles, the SAE has established a widely accepted classification of six levels of autonomy:
- Level 0: no driving automation, the driver is fully responsible for driving (steering, acceleration, braking, parking and any other manoeuvre).
- Level 1: driver assistance, with support systems such as adaptive cruise control. The driver remains responsible for driving and must be ready to take control at any moment.
- Level 2: partial driving automation, with advanced driver assistance systems that can take over steering, acceleration and braking in specific situations, such as motorway driving assistance. The driver must stay vigilant and actively supervise the technology.
- Level 3: conditional driving automation, which calls upon driver assistance systems and artificial intelligence to make decisions based on the surroundings around the vehicle. The driver must no longer supervise the technology, but must be present, alert and able to take control of the vehicle at any time.
- Level 4: high driving automation, does not require any human interaction in the vehicle’s operation because it is programmed to stop itself in the event of system failure.
- Level 5: full driving automation, which means that the vehicle can drive itself anywhere and under any conditions without any human interaction. Therefore, no need for a steering wheel or pedals.
The 360Lab, dedicated to intelligent mobility
The 360Lab of the SnT of the University of Luxembourg is the first thematic research laboratory focusing on smart mobility. The purpose of the 360Lab is to serve as an umbrella for research projects sharing common equipment and complementary expertise to conduct strategic and collaborative research in the broader area of mobility innovation.
The fields of research of 360Lab include: autonomous systems, vehicle-to-everything communications (V2X), artificial intelligence for automated mobility, safety and resilience of automotive systems, mobility sensing, modelling and simulation technologies, as well as transport planning.