News

How automated cars and infrastructure work together
Presentation of MEC-View results

Ulm University

City traffic can be pretty confusing: pedestrians hidden by cars, cyclists crossing the path, buses suddenly re-entering traffic. The project „MEC-View“ is researching how street lights can make traffic safer and provide automated vehicles with an overview of the traffic situation. In order for this to work, video and lidar sensors have to be mounted onto the lanterns. Using mobile radio technology, the sensors provide vehicles with important information in real time, to enable them to quickly and reliably recognise obstacles – whether they be other cars, bikes or pedestrians.

Following more than three years of development, the project results are now being presented. Partners in the project, which has received 5.5 million euros in funding from the Federal Ministry for Economic Affairs and Energy (Bundesministerium für Wirtschaft und Energie, BMWi), are Bosch (consortium leader), Mercedes-Benz, Nokia, Osram, TomTom, the company IT Designers and the universities of Duisburg-Essen and Ulm. The City of Ulm is an associated partner. Over the last three years, researchers have been testing the sensors on the street lights and the network technology in Ulm. The knowledge gained from this project will now serve to further develop car technology, automated driving and mobile radio technology. What’s more, the infrastructure that has been set up for this project can now be used for further research projects as well.

A bird’s eye perspective yields a better overview

At a height of six metres, street lights tower over traffic. They offer the perfect perspective for recognising what is happening, for example, at busy intersections, which is something that automated vehicles will need in the future. The built-in sensor systems in vehicles with cameras, radar and lidar sensors do already enable a precise 360-degree panorama view. However, from a driver’s perspective, it is not always easy to see if a pedestrian is hidden by a truck, if another car is coming from a street that is not very visible, or if a cyclist is approaching from the back and abruptly changing lanes. “Since the car itself cannot look around the corner or through the walls of buildings, we use the sensors on street lights to expand the vehicle sensor range of detection”, explains Dr Rüdiger Walter Henn, head of the “MEC-View” project at Bosch. The project partners have developed the corresponding hardware and software to process the images and signals of infrastructure sensors, combine them with high-resolution digital maps (HD maps) and transmit them to the vehicle via mobile radio. There, the data is merged with the data from the vehicle’s sensors to create a precise picture of the situation with all the relevant traffic participants.

Data transmission via mobile radio

Modern mobile radio technology enables the transmission of sensor information with minimal latency. While LTE mobile radio technology with an optimised configuration was used for the project “MEC-View”, real-time data transmission is a basic function with the new communication standard 5G. The central task of latency-optimised mobile radio is not only the virtually delay-free transmission of data via radio signals, but also its processing as close to the source as possible. This task is being completed by special computers, so-called Mobile Edge Computing servers, or MEC servers for short, which are directly integrated into the mobile radio network. They combine the sensor data of the street lights with those of the vehicle’s environment sensors and high-precision digital maps. In this way, they create a local environment model with all of the available information on the current traffic situation and make it available to the vehicles via mobile radio. In the future, for example, the traffic control centres within cities could be equipped with such servers, in order to share all the data with all of the traffic participants, regardless of manufacturer.

Seamless merging onto a main road

In Ulm, project partners have been testing the interaction between automated driving prototypes and infrastructure sensors in real traffic situations since 2018. At an intersection in Ulm-Lehr, the street lights have been equipped with sensors for this purpose. The vehicles approach the intersection in an area with limited visibility on a side street, for example, and then merge into traffic on the main road. Thanks to the newly-developed technology, the automated driving prototype now recognises the traffic participants at an early stage and can adapt its driving strategy accordingly. In this way, the vehicle detects gaps in traffic on the main road and merges seamlessly, without stopping. This not only makes city traffic safer, but also improves the flow of traffic. The infrastructure set up for the project will remain in Ulm and will be available for use in further research projects.

Only by transferring the environment model to the automated vehicles does the hidden become visible

The involvement of the Institute of Measurement, Control and Microtechnology (Institut für Mess-, Regel- und Mikrotechnik, MRM) at Ulm University in the MEC-View project has primarily included infrastructure and an automated test vehicle. Specifically, the engineers and computer scientists operated some of the sensors that are mounted at the test intersection in Ulm. They also developed the algorithm for compiling the sensor data (fusion) into an environment model of the intersection on the MEC server. “One of the core components of the system is the low-latency fusion of the data from different sensors, ie the fusion carried out with the shortest possible delay, to yield a precise overall image of the traffic situation”, explains Dr Michael Buchholz, who coordinated the involvement at the MRM.  Transferring the environment model to the networked and automated vehicles allows areas that are hidden from a car’s own sensors to become “visible”.

The MRM institute, headed by Prof Klaus Dietmayer, also provided one of the automated test vehicles. In the MEC-View project, the vehicles were able to carry out the complex process of merging into traffic in an automated fashion. The engineers and computer scientists at Ulm University developed and applied a planning algorithm for this purpose. This algorithm combines the sensor information from the individual car and the transferred environment model, enabling the test vehicle to merge into traffic either before or after another car, or even into a gap on the main road if the situation allows. “This procedure checks the reliability of the information provided by the infrastructure. In addition, it takes into account the uncertainties of the data during planning”, Dr Buchholz adds. The researchers tested and evaluated the developed procedure at a pilot station, the test intersection in Ulm-Lehr.

The results are presented online on the website www.mec-view.de, where presentations, films, pictures and explanatory texts about the research project can be found.

Text: Caroline Schulke (Bosch) / Dr. Michael Buchholz & Annika Bingmann (Ulm University)

[Translate to english:] LiDAR- und Kamerasensoren an der Pilotanlage
[Translate to english:] LiDAR- und Kamerasensoren an der Pilotanlage (Bildquelle Uni Ulm MRM)
[Translate to english:] Sensoren an der Pilotanlage
[Translate to english:] Sensoren an der Pilotanlage in Ulm Lehr (Bildquelle Uni Ulm MRM)
[Translate to english:] Trajektionsplanung und Umgebungsmodell
[Translate to english:] Links: Planung (bunter Pfeil) des automatisierten Fahrzeugs zum Einfädeln; Rechts: Modell der Umgebung im automatisierten Fahrzeug mit Karte, Lidar-Daten des Fahrzeugs (Punkte) und Objekte aus dem Infrastruktur-Umgebungsmodells (Boxen) (Bildquelle Uni Ulm MRM, MEC-View)
[Translate to english:] Automatisiertes Versuchsfahrzeug
[Translate to english:] Automatisiertes Versuchsfahrzeug (Bildquelle Uni Ulm MRM)