Predictive Autonomy Lab

Assistance systems in which vehicles take over all or part of the driving task in certain situations offer us convenience and perhaps even time for other activities. Whether and to what extent this has an impact on safety and how such systems should ideally communicate and cooperate with humans is still largely unexplored.

With the Predictive Autonomy Lab (PAL), Fraunhofer IESE offers companies from the automotive industry a controlled environment for investigating such effects.

The laboratory is used for empirical driving behavior research from a safety perspective. The influences of a human driver's environment on their driving behavior can be investigated here. To this end, the scientists at Fraunhofer IESE conduct driving simulator studies with appropriate sensor technology in order to understand the causal behavioral decisions of the test subjects. Examples of this include monitoring systems such as eye trackers and the physiological observation of vital data.

The focus of the driving simulator is on understanding human behavior in various safety-critical driving situations. This enables – together with suitable methods and tools – the generation of scientifically validated, situation-sensitive and safe driving models. The driving models can be used profitably in safety validation as part of highly automated driving systems and advanced driver assistance systems in the automotive industry. On the one hand, this allows real-to-sim gaps to be minimized. The aim is to obtain reliable and valid simulation results that can be used for decision-making in practice. On the other hand, the driving models can be used to reduce unnecessary safety interventions in safety monitoring systems.

Fraunhofer IESE offers infrastructure services in the context of safety and driving behavior research: Companies from the automotive industry can use not only the technical environment, but also the expertise from numerous scientific research projects, e.g. to integrate situation-sensitive knowledge into their driving systems, to have driving strategies independently evaluated by Fraunhofer, or to evaluate new system concepts.

Research services/investigations for the automotive industry

• Checking whether assumptions in the safety concept, e.g. for people taking over the driving task again, are valid
• Evaluation of which driving situations are particularly difficult for people to master after they have turned their attention to other things
• Development of digital twins of human drivers for innovative autonomous driving systems as part of SAE Level 2 ADAS (Advanced Driver Assistance System) to improve safety-relevant driver monitoring
• Development of digital twins of human drivers for innovative autonomous driving systems within the framework of SAE Level 3+ ADS to improve the safety-relevant driving decisions of the ADS in mixed traffic situations
• Minimization of the real-to-sim gap in the context of virtual validation
• Research into the perceived driving risk of passengers in SAE Level 3+ ADS

Technical features

• Realistic design mock-up with motion system (15cm stroke) for generating motion cues
• 180° panoramic projection of the virtual environment over a width of >3m
• Digital exterior and rear view mirrors
• Large central display with touchscreen function, e.g. for displaying infotainment systems or for use as a human-machine interface in studies
• Realistic 5.1 surround audio in the vehicle mock-up to create an immersive world
• Simulation software tailored to research studies with a focus on the interaction of the test person in the scenarios
• Behavioral research software for time-synchronous data fusion
• Sensor technology for behavioral research, such as eye trackers and physiological vital signs sensors