Traffic situation in which the automated bus can "look around the corner" © DAI-Labor

14 September 2021

BeIntelli lets the people of Berlin experience the mobility of tomorrow.

Traffic jams on the way to work in the morning and crowded buses at rush hour could soon be distant memories. Instead, the mobility of tomorrow will be “characterized by different intelligent means of transport: vehicles, and that doesn’t just include cars, but also vans, delivery robots and autonomous buses, that will work intelligently like human drivers, will be sustainable and carbon-neutral, and will be adaptable to individual needs,” predicts Prof. Şahin Albayrak, founder and director of the Distributed Artificial Intelligence Laboratory (DAI-Lab) at the Technische Universität Berlin, looking to the future with optimism. This computer scientist is convinced that autonomous driving will make traffic safer and make it flow better, as well as allowing a better combination of different modes of transport and optimizing utilization of different types of vehicles depending on our needs. “In the future our needs will be the most important factor, not owning our own vehicle,” Albayrak adds. “This will make transport more sustainable and more efficient—we’ll need less parking space, cities will be worth living in again, and rural areas will be well-connected for all scenarios in logistics and personal mobility.”

The potential of autonomous driving for society, security, and Europe as a business hub is enormous, something the German government is also banking on: Vehicles with autonomous functions are to be brought into regular operation by 2022, which will make Germany the first state in the world to bring driverless vehicles from the field of research into our day-to-day lives. But not all citizens are quite as convinced of the benefits of autonomous driving as the expert from the Technische Universität (TU) is. According to surveys, 45 per cent of car drivers don’t believe that this vehicle technology is reliable or are concerned about hackers. For Prof. Şahin Albayrak, this reserve has a simple explanation: “Very few people have had the chance so far to see autonomous vehicles in real life or to actually get into one and really grasp the technology.” Changing that is one of the aims of the “BeIntelli” project led by this entrepreneur and founder of various start-ups: “Our aim [is] to make it possible to experience the mobility of the future and to encourage acceptance and positive change by breaking down people’s misgivings,” he explains. “With BeIntelli, we want to be transparent with people, invite them to try it out, and get them interested in the mobility of the future.” To do this, BeIntelli is building various touch points. Alongside the website as a digital port of call, a showcase is also planned to offer information and events on the topic of autonomous driving. “To get citizens more interested, we’re looking for proactive exchange wherever that’s possible, like at an exhibition as part of Knowledge City Berlin, plus other events and conferences as the year goes on,” Albayrak says.

BeIntelli is bringing current research to the streets

Principally, though, BeIntelli is bringing a fleet of different vehicles, from cars to vans, as visual models to an urban testing site at the heart of Berlin. The cars can demonstrate how personal transport will work, while the vans will show different scenarios in autonomous package delivery and logistics. But looking alone isn’t enough: “In particular the bus, that’s currently undergoing some modifications, will act as a mobile field test,” according to the director of the DAI-Lab. “Citizens can get on the bus and take a look at displays that show everything the vehicle senses and how it responds to the environment around it.” It’s through this targeted testing in genuine test environments and through public demonstrations that visibility is to be increased for society, business, politics, and science—like a kind of shop window—as the website reports. This will allow users to learn how to interact with autonomous vehicles, as well as identifying more fields of action for the new technology at this early stage in its development. The scenarios are based in urban space to start with, however they can be “adapted and transferred to rural spaces, too,” Albayrak points out.

Map with the test track in the heart of Berlin © DAI-Labor

From DigiNet PS to AI Mobility OS by BeIntelli

BeIntelli is not the first project by the DAI-Lab at the TU Berlin to bring the mobility of tomorrow to the Berlin of today. For two years, scientists collaborated with specialists from the fields of research, business, ICT, and transport along a test route on Straße des 17. Juni to gather data on automated and networked driving. “BeIntelli is based on the findings from this initial project, which was called DigiNet-PS (hyperlink to: https://ki-berlin.de/en/blog/article/artificial-intelligence-creates-the-mobility-of-the-future),” Albayrak goes on. The core element of the project, comprising a consortium of twelve partners and funded by a €12.91m grant from the Federal Ministry of Transport and Digital Infrastructure, is the development of a scalable software stack called AI Mobility OS; “A kind of operating system allowing different autonomous modes of transport to move around intelligently. The software stack equips the individual vehicles, the cloud, and even infrastructure with a range of basic capabilities.” For this concept, called “distributed intelligence”, the test route—which runs from the Brandenburg Gate via Ernst-Reuter-Platz to the Gedächtniskirche—will be divided into approximately 400-meter segments and each one assigned its own computing unit. The stack as a whole will consist of various levels: hardware, AI middleware, ADAS++, and an AI platform. Depending on the hardware, the AI middleware is responsible for integrating devices; collecting, transforming, aggregating, and validating data; for synchronizing data between vehicles, street segments, and the cloud; for data security and subsystems; and for controlling the devices. The ADAS++ level, meanwhile, takes over the analysis, modeling, and forecasting of the data, as well as making decisions for the respective subsystem based on these. In this way the AI Mobility OS makes it possible to use artificial intelligence on different levels, from using raw data like weather and parking occupancy all the way to applying this to pooled data from individual sites, like adjusting traffic lights to different weather conditions such as slippery roads or altering routes to avoid traffic jams, noise pollution, and accidents. “The AI platform comprises repositories and tools based on the AI middleware and ADAS++ to provide different data, AI models, services, and applications, which different participants in the platform economy can use,” the leader of the consortium explains.

Prof. Sahin Albayrak © DAI-Labor

Platform economy to develop real innovation

However, BeIntelli doesn’t just make the real-time raw data available to the twelve members of the consortium who are supporting the project in different aspects including software development, kitting the test vehicles out with autonomous driving technology, and providing their consulting expertise. Other market participants including start-ups can also use the data and models to develop and test out innovative applications in a real environment.It’s already been proven in other domains that platform economy promotes the development of genuine innovations,” the long-serving TU professor says. “We’re convinced that platform economy forms the foundation to develop new solutions for the mobility of the future. It’s the basis for new ecosystems since it brings together the different stakeholders (platform operators, providers, and users) and makes it possible to exchange ideas and services.” By the time the project ends in June 2023, BeIntelli aims to contribute to an ecosystem around the mobility of the future that is as vibrant as possible. Plus, “we want to realize a reference configuration of different digitized vehicles of the future, as well as implementing the software stack as an AI Mobility OS in different modes of transport (test vehicles) and infrastructure,” Prof. Dr. Şahin Albayrak hopes, adding: “At the same time we’d like to introduce these topics for the future to mainstream society and encourage acceptance of them.” After all, the automation of Germany’s roads will be a reality by 2022 at the latest. Now what’s important is trusting the technology and making the most of its potential.