Fraunhofer Institute for Industrial Engineering IAOCompanies that want to harness the potential of new technologies usually require some form of support. Fraunhofer IAO is providing help within the framework of the “Exploring Projects” initiative. For example, the development of AI applications for the logistics company Zeitfracht.
From search machines to chatbots to spam filters: In today’s virtual world, we are constantly interacting with applications that work on the basis of artificial intelligence (AI). Moreover, the possibilities offered by such technology are increasing in the real world too — autonomous vehicles are driving on the roads, robot vacuums are cleaning houses and machines are being increasingly used in agriculture to decide when to water or harvest a field.
So what potential does this hold for German SMEs? What advantages does this future technology offer companies from the fields of industry or service? Many companies in Germany have yet to find a concrete answer to these questions. “Smaller companies in particular often lack the opportunity to identify and implement AI use cases,” says Dr. Matthias Peissner. As institute director, he heads the Human-Technology Interaction research unit alongside his role as joint head of the AI Innovation Center, a joint initiative of Fraunhofer IAO and IPA at the Institute Center Stuttgart. Here, Peissner wants to give companies from Baden-Württemberg the opportunity to become familiar with AI.
Testing AI applications
“Exploring Projects” is the name of the initiative whereby Fraunhofer experts work with companies to make new technologies useable. The associated projects are funded by the Baden-Württemberg State Ministry of Economic Affairs, Labor and Tourism. These include the AI Innovation Center’s projects in which specific AI applications are developed and tested in companies.
Participation is free for companies and a summary of the project results is published in return. However, companies do not need to worry about disclosing too much information: “We design feasibility studies on specific issues, but we only publish the results that can be transferred in a similar fashion to use cases in other companies,” explains Peissner. At the end, there is knowledge transfer — without the need to invest a lot of money in research and development.
One Exploring Project with the logistics company Zeitfracht shows what sort of problems are being addressed. Zeitfracht, which is active in the wholesale market, among others, was looking for a prediction model that would enable it to determine at any given point the right inventory levels for numerous products. “The aim was to minimize warehouse costs while still ensuring fast delivery,” explains Peissner.
Zeitfracht previously used a prediction model from the field of supply chain management for this purpose. “We wanted to find out whether we could achieve better results with the help of machine learning,” says Peissner. The result: a firm “yes,” as the AI expert reports. “For the majority of the year, our algorithm was more precise than the previous model.” According to Peissner, it was only during high trading periods, such as close to Christmas, that the employees’ experience and the existing process had the edge, providing more nuanced forecasts. However, this was a result that the scientist could happily live with. “Our objective is human-centered AI, i.e., technology where the decisions are still made by a person,” he says. In this scenario, the algorithm’s advantages can be combined with the employees’ expertise in the best way possible.
Free project start for companies
Since the start of the initiative in the fall of 2019, the AI Innovation Center has completed a total of 21 Exploring Projects. Demand is so high that almost every second application has to be rejected. “Selected projects are ones that are both innovative and technically feasible. Furthermore, they must offer the greatest possible economic benefit — taking into account the widespread impact beyond the applicant company too,” explains Peissner. In addition, the measures must be implemented within an appropriate time frame. “We usually carry out a one-month ‘Quick Check’ to start with and if the result is positive, we then invest three months in the Exploring Project,” he says. Afterwards, the companies can continue to use the developed solutions. “However, this then becomes subject to a charge because the German state government funding only covers the proof of concept,” says Peissner. Participation is attractive nevertheless. “This is because at the end of the Exploring Project, they know for definite whether it is worth paying the money.”
Exploring Projects: Looking for support?
Fraunhofer IAO supports companies and public institutions with innovation processes. As part of the “Exploring Projects” initiative (prototypical implementation), researchers are supporting their partner institutions by teaching them about new technologies and making these accessible to them. These are always provided for a certain period. Here you can find a selection of our Exploring Projects that are currently offered.