The global crisis situations of recent years have repeatedly put the resilience of the local supply systems to the test. The Corona crisis highlighted the fragility of the economy and its dependence on global production and supply chains. The impact of armed conflict, such as the war in Ukraine, on food and energy supplies is felt throughout the world. But how can a critical supply situation be prevented from becoming catastrophic?
In the ResKriVer project, a digital platform is being implemented that uses artificial intelligence (AI) applications to collect, create and communicate crisis-relevant information for supply chains. In particular, it aims to forecast the impact of bottlenecks in the supply chains of companies and public demanders. We spoke with project leader Professor Thomas Hoppe of Fraunhofer FOKUS and HTW as well as Christoph Stiller, subproject leader of the Berlin Fire Department, about the genesis of the ResKriVer project, the potential for upcoming crises and the transferability of the platform.
Prof. Hoppe, please explain how the ResKriVer project came about and which project partners are on board?
Prof. Thomas Hoppe: Through an announcement from Berlin Partner, I learned about an application workshop for the call for proposals at the time. At short notice, I submitted the pitch presentation of an idea on how we want to support the collection and secure storage of supply chain information and its analysis through a distributed platform. As a result, the Berlin Fire Department, rbb, Merantix, FhG IML, HFC and YOUSE then contacted me saying that they were interested in participating. The rbb then made contact with Charité, the fire department with vfdb and KomRe, and I with Condat and eccenca.
How does your digital platform work and what influence does AI have here?
Prof. Thomas Hoppe: I can't explain here - yet - exactly how the digital platform works. The basic idea behind it is to use semantic modeling - and this is common to at least four of the AI crisis management projects - to describe the respective application area with methods of symbolic AI. On the one hand, this allows easier adaptations and extensions to new crisis/disaster scenarios; on the other hand, especially for crises and disasters, large amounts of data are not available to use non-symbolic AI methods, which include machine learning, artificial neural networks and deep learning, on a large scale.
In the context of ResKriVer, we are using different AI methods for this purpose: speech processing methods for analyzing social media postings, Bayesian networks for analyzing the weak points of supply chains, and reinforcement learning methods for optimizing simulations. We also plan to use Bayesian networks to identify high-risk patients in need of urgent treatment in the event of a crisis, as well as learning methods to compress sensor data collected by drones.
Prof Thomas Hoppe © Berlin Partner
We have been crisis-tested in the last two years, especially due to the Corona pandemic, and have had to experience various bottlenecks due to the shutdown of productions. What systemic weaknesses could ResKriVer help with in the future?
Prof. Thomas Hoppe: As far as supply chains or, more broadly, value chains are concerned, our goal is, on the one hand, to identify weak points, such as "bottlenecks" that lead to disruption in the event of disruptions, and alternatives, such as alternative suppliers, producers or substituting products. And on the other hand, to use real data to make predictions about delivery times, delays and available quantity frameworks.
However, we face a major problem here concerning the availability of the required data. In the globalized world economy with its distributed production and supply processes, we have so far only been able to record sections of these chains - which are actually networks. Although we are developing technologies for the distributed, secure storage of confidential information and for analysis and simulation, we are not yet able to access the necessary data. We need contacts and information on the entire chain, from raw materials to producers, transport companies, suppliers and end customers, in order to make these weak points visible and transparent.
Mr. Stiller, the focus of the Berlin Fire Department's subproject is on ensuring the ability of authorities and organizations with security tasks (BOS) to act. What does that mean in the case of the fire department?
Christoph Stiller: As the Berlin Fire Department, we have to be and remain operational at all times in all areas - including firefighting, emergency rescue and technical assistance, for example. To do this, we need the appropriate equipment: protective gear for the emergency services, technical equipment and medical material to care for patients.
Especially in crises, we depend on functioning supply chains. This includes not only goods and resources, but also crisis-proof communication channels. In the ResKriVer project, the Berlin Fire Department, as a practical partner, provides the expertise for supply and communication processes for authorities and organizations with security tasks.
What crisis scenarios do you encounter regularly or unexpectedly - such as the Corona pandemic? How can the ResKriVer platform help to increase resilience in general?
Christoph Stiller: Examples of unforeseen crises that the Berlin Fire Department has dealt with in the past include the power outage in Köpenick, hurricanes Ylenia and Zeynep in February 2022, or the vegetation fire in Grunewald in July 2022. With some regularity, for example, the Berlin Fire Department provides support during bomb defusions or cares for an increased number of patients during scenarios such as extreme heat or major sporting events. For example, the 2024 European Football Championship, with matches in Berlin, will challenge the Berlin Fire Department in all areas.
ResKriVer contributes to a general increase in resilience by making information accessible quickly and consistently. This includes information such as stockpiles and other available resources, information about impassable areas accessed by drones, and information from the population gathered through social media monitoring. A robust information base is essential for decision-making, and not just in crisis scenarios. A big question at the moment is how to maintain communication channels for key information in the event of a power outage. But also in the everyday life of a BOS, this information basis can strengthen the resilience of an organization by identifying later potential challenges at an early stage and making processes in one's own organization transparent and communicable.
Christoph Stiller © Berlin Partner
Let's stay with the topic of communication: Addressing target groups appropriately is essential for overcoming crises. What adjustments do you see here with regard to the work of authorities and organizations like yours?
Christoph Stiller: The Berlin Fire Department currently communicates via various media, some with different objectives and target groups. For example, Instagram is used primarily to recruit personnel.
One possibility being pursued in the ResKriVer project is highly target-group-oriented communication in locally limited emergency situations, such as a power outage in a particular district or a local smoke situation. Using the appropriate technical tools, messages could be scattered in such a way that citizens in the vicinity of the damage site are explicitly informed. At this point, cooperation with the other project partners is central. In this context, rbb is developing a "crisis compass." In order for media institutions, such as rbb, to be able to provide the population with the right information in the event of a crisis, crisis-proof communication channels are needed between media institutions and authorities and organizations with security tasks.
The general approach of the project is transferability to a wide variety of crises. To what extent is this possible and which scenarios are being considered here?
Prof. Thomas Hoppe: In the project, we focus on regional or national, longer-lasting crises. Here, our focus is on large-scale fires, the supply of blood reserves, treatment planning for elective risk patients, and communication with the population and in blackout situations. Each of these scenarios is representative of an entire class of situations.
Large-scale fires, for example, are characterized by the fact that it is not easy to get into the affected areas to gather information and that emergency forces are tied up for a long period of time and have to be supplied and replaced. In terms of character, this is transferable to flood situations, in some cases to drought situations or situations in which traffic routes are no longer available throughout the area (e.g. terrorist acts or war situations).
The supply of blood is a problem that arises in scenarios where there are large numbers of injured patients, but it also becomes acute in the summer months during the travel season.
The treatment planning of high-risk patients who do not yet require immediate treatment is something we learned about at the beginning of the Corona pandemic, where non-urgent interventions had to be postponed for the time being. Here, we want to develop an exemplary solution that is effective in pandemic situations as well as in cases where emergency rooms and intensive care units reach their capacity limits due to many acute cases.
From the perspective of a computer scientist or information scientist, we can achieve the transferability of these approaches through knowledge-based systems, i.e. systems in which knowledge about the application areas and scenarios is explicitly represented and thus interchangeable, but the processing methods themselves remain independent of the application areas. For learned models this is only possible to a very limited extent.
The ResKriVer project brings together technology developers, application partners and enablers. Were you able to learn from the use cases of other partners which could be relevant for the work of the fire department in crisis situations?
Christoph Stiller: Like the Berlin Fire Department, the other partners in the field contribute specific use cases to the project. There are also overlaps here, such as in the area of communication with the population, support for the work of crisis teams or the tracking of supply chains of crisis-relevant goods. The knowledge and experience of the respective practice partners are incorporated into demonstrators, which in turn can benefit various stakeholders. For example, the project will extend the fail-safe communications network that KomRe is contributing to the rbb. This expansion of the communications network also represents a benefit for the Berlin Fire Department, because it increases the range of incoming and outgoing messages in the event of a power outage.
What is the significance of the project for Berlin and can it also serve as a model for other cities and federal states?
Prof. Thomas Hoppe: The transferability of the project results to other regions and federal states is a goal for us. Here, through rbb for the public broadcasters, through vfdb e.V. and the Berlin Fire Department, and through Charité as the largest full-service university hospital, we have good opportunities to serve as a role model and to transfer the solutions of the project.
Prof. Hoppe, with the conference „Vernetztes Krisenmanagement 2022“ in October, you were able to bring together users, decision-makers and scientists involved in civil hazard prevention. What is your conclusion of the four-day conference? What insights did you gain?
Prof. Thomas Hoppe: For this first conference "Networked Crisis Management 2022" with around 127 participants, the focus for us was on networking five of the six AI crisis management projects. This was also very much welcomed by the „DLR Projektträger" (DLR Project Management Agency) and the BMWK. In the preliminary planning stages, we consortium leaders of these projects have already learned that our projects are not in competition with each other, even if there are certainly overlaps, but actually complement each other. Therefore, I think we have taken another step towards mutual networking with this event. Our goal here is to use a common use case to show how the project results can interact. For the next conference, which is scheduled to take place at the end of September 2023 at our Fraunhofer Institute FOKUS in Presence, I have definitely learned that we should start planning earlier, ignore all uncertainties due to Corona, and make our initial results transparent.