LOKI

In the project Collaboration of Aviation Operators and AI Systems (LOKI), we analyse approaches to collaboration between humans and AI systems. An important building block for this is the investigation of metrics for state detection of the human partners. In the project, we develop prototypes of domain-specific AI systems, such as the digital co-pilot, and use them to develop guidelines for the design of the interface between users and AI systems.

The following DLR institutes and external project partners are involved:

· Institute of Aerospace Medicine
· Institute of Flight Guidance
· Institute for AI Safety and Security
· Institute of Data Science
· Institute for Software Technology
· Institute of Air Transport

External partners of the LOKI project:

· DFS Deutsche Flugsicherung GmbH
· University of Lübeck, Department of Engineering Psychology
· Austro Control GmbH
· German Aviation Association (BDL)

Contribution Institute for AI Safety and Security

The Institute for AI Security is involved in four of the five main work packages and is investigating, among other things, the requirements for trustworthy and transparent AI for the use of digital co-pilots. The project will run for four years from 2022 to 2026.

In the future, people will work together with artificial intelligence systems. Interaction with AI systems will characterise activities in a wide range of fields such as aviation, medicine and space travel, as well as in everyday life.

When designing AI-based systems, the psychological requirements for human operators and the human-AI interface must be analysed at an early stage and taken into account with appropriate design solutions. In contrast to existing forms of assistance, AI systems can learn and thus successively adapt to the environment. By integrating learning algorithms into existing adaptive assistance systems, the system function of adaptivity to human states can be improved. In this way, the purpose of the system - to support the acting human being in his or her work task - will be achieved more effectively. Learning algorithms can identify patterns in complex biometric data sets that arise and are captured during the performance of activities. These patterns are mapped through training to a particular expression of a categorical human state variable. Such a mapping then provides the assistance system with more precise information about the type of assistance currently required.

With this project, we want to develop guidelines for a human-centered design of communication, but also collaboration between users and AI systems. Our focus is on areas of activity in air traffic management in which operators work together collaboratively. The focus of our project is on humans as users of AI systems and the question of how acceptance, comprehensibility, satisfaction and predictability can be ensured for the users of AI systems.

The project team works in an interdisciplinary and cross-cutting way. LOKI is supported by DFS Deutsche Flugsicherung GmbH, Austro Control GmbH and the BDL, Bund Deutscher Luftfahrtwirtschaft, as well as scientists from the Institute for Multimedia and Interactive Systems (IMIS) at the University of Lübeck as external partners.

The objectives of the project are: to analyse approaches to human-AI systems collaboration, to investigate metrics for human partner state detection, to develop prototypes of domain-specific AI systems, and to develop guidelines for the design of the user-AI system interface.