Getting Support for New and Familiar

January 23rd, 2023 | by Andreas Richter

(8 min read)

At the and of a year or with the begin of a new one in most regions of the modern world you will get presents. That’s why we had a look into the German research funding of automated driving topics; we were driven by our impression that something might be odd as we have recently mentioned in our forecast or some re-invention might be subsidised.

In November 2022 there was the conference about research and technology for automated driving hosted by the Federal Ministry for Economic Affairs and Climate Action of Germany. In the context of this conference a collection of currently funded research projects was published and we have reviewed it.

Three Federal Ministries of Germany are supporting the majority of the German research projects focusing on automated driving topics:

  • Federal Ministry for Economic Affairs and Climate Action (BMWK): 47 projects primarily with focus on mobile communication technologies, eco-system platforms, artificial intelligence approaches and validation
  • Federal Ministry for Research and Education (BMBF): 31 projects mostly with focus on embedded components such as electronic control units and engines, sensor setups, mobile communication technologies, artificial intelligence approaches
  • Federal Ministry for Digital and Transport (BMDV): 15 projects mainly with focus on testing areas, vehicle concepts and artificial intelligence approaches

Reads familiar? Yes, there are overlaps and interesting as well as interesting projects among them.

Artificial Intelligence for everything

Machine learning is expected to solve everything that is too complicated for rules. Often, pattern recognition is applied or, at least, we believe that the networks recognize patterns. In research sometimes such networks are able to find new solutions which humans haven’t considered yet, because it was too much outside the box. But most of the AI projects are just applying AI for doing something such as road user detection, interior surveillance, improved detection during bad weather, improving vehicle dynamics, improving traffic management, improving communication between traffic participants, retrieving scenarios from vehicle and traffic data, retrieving vehicle conditions from vehicle and traffic data, supporting, the development of electronic and computing components and for automating the operations of mobility as a service.

In general, a large quantity of training data have to be applied to successfully use AI in pattern recognition applications. Hopefully, these data cover every characteristic of the problem to be solved otherwise the neural network will overfit. To solve this problem, there are also funded projects to collect training data and provide them together with models via platforms. Unfortunately, the number of these projects is small and they are focusing on sensor data which will be provided by the project partners… which means that, again, the data will be only a snippet of the real-world conditions and limited to the sensors the data was collected with.

But there are also some really interesting projects: These try to make AI-based object recognition ASPICE compliant or try to integrate knowledge (rules, guidelines, schemas, etc.) to improve recognition and make it more reliable and explainable. This is definitely added value from which, hopefully, a lot of “let’s apply AI to do something” projects will benefit and deliver more than just a proof of concept.


By the way: reliable. Everything that was developed and proved to be useful has also been proven to be safe (and secure – in German you only have to prove it “sicher”). Therefore, a lot of funded projects work on design, development, test, approval, operation and monitoring processes for components, systems and their application. In the focus regarding automated driving is the validation and verification of the driving function. The scenario-based testing seems to be the commonly agreed solution but still is a mountain of work to do. Some projects try to deliver solutions for all sections of the lifecycle; some of them try to focus on one specific section. Hopefully the concepts will fit together afterwards.

Platforms for everyone

Another hot topic is the development of platforms for providing and connecting stuff. Especially the Gaia-X projects (funded by BMWK) want to develop a federation of data infrastructure and service providers. This includes a platform to connect communication and management of traffic and traffic infrastructure and fleet operation, training data and services for AI application (see above) and identification infrastructure for mobility data and production data. The latter is also funded by BMBF because two are better than one?! BMDV only focuses on a platform to manage traffic (to let it form an emergence corridor).

It is interesting to see that these huge projects (project budgets between 19 and 27 million EUR with often very similar big project partners – of which more later) want to create a new eco system of (base) services and data. Interesting because barely any project partner will ever set up mobility services or provide and operate larger parts of mobility services. There are no OEMs, no fleet operators nor any mobility service providers represented in the project consortia. Once can assume that they will build their own platforms.

Test areas and vehicle concepts

Also developing and operating test areas (or “test fields” as they are sometimes called) is still a possibility to get funding. Germany had quite some test areas such as the motorway A9, another one in the region of Braunschweig, in the city of Kassel or Dusseldorf or Berlin or Karlsruhe. Some of them get continued funding for testing “something”. Often, only the operator of the test area deploys a few test campaigns, and the interest by third parties to use a test area is often very low. But again, having multiple is better than one and, therefore, there is still tax payers money being spent to test how low-speed shuttles can be operated, made more reliable and a service of such shuttles could be set up (you can read more about this in our feature about the lost generation of automated driving).

In addition to test areas also vehicle architectures and concept developments are still getting funded. Some of them are just focusing on electronic components and how to make them more energy efficient or create new general designs but some of the projects want to create complete new modular vehicle platforms. A 32 million EUR funded projects of universities and some tool suppliers together with one Tier-1 want to setup and demonstrate in some of the test areas a modular and scalable vehicle platform with focus on connectivity to road and cloud infrastructure as well as sensors and updatable soft- and hardware. Thus, develop a little of each. The project wants to provide a substantial contribution to the interdisciplinary innovation power of German research. At least they were able to combine the largest number of buzzwords in their description without being specific, but still the project provides a lot of topics for students to work on papers and theses.

One more thing

Most of the projects have an expectable scope and that’s why we have clustered them. But sometimes a project is surprising with something interesting – with respect to something new or something that we thought was already solved. For example:

  • There is a project funded, which is working on human-centric highway lane-keeping and -changing system (SAE level 2) for commercial vehicles with only on Tier-1 and no OEM.
  • There is an undertaking that would like to adopt organic computing methods for self-healing of electronic control units.
  • A project consortium of universities, public authorities and traffic enterprise wants to improve robo-shuttles to go faster than 50 km/h and the project consortia lacks a Tier-1 as well as an OEM.
  • There is a project to develop and test ground-penetration radar for localization despite the fact that this was proved years ago and some startups are already offering this as a product.
  • One consortium works on the improvement of passive radar reflection of vulnerable road users with potential identification (hopefully, they cover the data privacy issues as well).
  • The traffic operator of the city of Hanover, together with the road infrastructure supplier and traffic management managers, tests whether vehicle-to-anything-technology could be used for public transport priority at traffic lights. Well, that is a really interesting research question!
  • Some project states that they want to define and develop “standard-like” interfaces e.g., for backend components. We won’t ask why they are not contributing to real international standardization organizations?


Supporting something new is always a good idea because sometimes you have to spend a lot of money before you know if the new thing is really helpful. As we have mentioned recently in our forecast, Germany’s Mittelstand employs 58% of the workforce and represents more than 99% of all companies. If you look into the list of the funded research projects 78% of them are managed by OEMs, Tier-1s and technology suppliers listed on the stock exchange or large research organizations. Fortunately, project members are often small or mediums-sized enterprises (SMEs) such as tool providers or universities (which contribute to these projects because they have to prove third-party funding – which again is funny because it is also tax-payers money like the institutional funding). The question is whether the big industry players really need a funding for doing research on topics, which should be in their genuine interest to improve their products. If a company is not spending a certain amount of their revenue for research then the company won’t survive in the long run. Often, the provision of resources within big companies is so tense that additional budget can help to decide to contribute to such projects. In this case, funding is also kind of revenue.

You could also understand the funding as more related to the collaboration work itself, to establish an information flow and good contacts between industry and research and also to let brains drain from research to the industry. If you keep this in mind then it is less of a pity that projects that merely put some buzzwords together do often not deliver results which can be reused by others; at least they are used in the products of the project members and, as such, reach the broader public indirectly. For the platform goals, in particular, developing vehicle concepts as well as applying AI for various things will, most probably, not deliver breakthrough results. Nevertheless, we hope that the projects working on validation methods will generate more knowledge how to test and approve complex products such as automated driving systems in a reliable and manageable effort.

And because competition is good for business the three Federal Ministries of Germany are sometimes supporting the same ideas multiple times. Just to be on the safe side.

Add a Comment

Your email address will not be published. Required fields are marked *