Servicenavigation

Project Details

Back

Projekt Print View

Learning from Humans – Building for Humans

Subject Area Human Factors, Ergonomics, Human-Machine Systems
Human Cognitive and Systems Neuroscience
Term from 2020 to 2025
Project identifier Deutsche Forschungsgemeinschaft (DFG) - Project number 433524510
 
Final Report Year 2023

Final Report Abstract

A holistic approach is important for the design of Human-Cyber-Physical Systems (HCPS), which involves the collaboration between humans and machines. While the technical viability of these systems has garnered attention in the past, human supervision, support, and intervention remain crucial. The goal of this project between the Technical University of Munich, University of Oldenburg and the DLR was to develop co-operative HCPS, where humans and machines work together, leveraging each other’s strengths and compensating for their individual limitations. We focused on understanding three key factors: human perception and its limitations, human decision-making influenced by cognitive load, and trust in automation. To achieve these goals, highly synchronized experimental research was conducted in stateof-the-art simulation environments in the contributing labs. The research involved behavioral measurements, human state measurements, brain-sensing technology, human-machine interaction (HMI) engineering and various modeling approaches. Through cluster analysis, we were able to uncover naturally occurring driving strategies at intersections. Our results revealed that individuals with visual field loss tend to rely more heavily on extensive and early scanning as a compensatory strategy. We also discovered that the complexity of the driving situation significantly influences the effectiveness of compensatory strategies. In addition, we found that humans evaluate actions in the decisionmaking phase differently when they interact with an autonomous vehicle or a human driven vehicle. This was expressed in functional near-infrared spectroscopy (fNIRS) brain activation and partly in the behavioral tendencies. Moreover, we observed interactions between working memory load and decision-making at the brain-level. To better understand these interactions, we used a model-based approach and found that even simple control actions in driving likely share resources with working memory tasks. This highlights why cognitive load and driving performance may interact in some cases. Furthermore, we used the same model to make a priori predictions about interventions that manipulate cognitive load to avoid severe underload. We found that all interventions tested in a model introduce switching costs that assistive systems need to consider. Additionally, we developed an assistance concept designed to enhance scanning behaviors and facilitate hazard avoidance at intersections and on straight roads. With the early detection of safety-critical maneuvers advanced driver assistance systems could play a vital role in preventing accidents at intersections, ultimately leading to a decrease both in overall accident incidence and potentially even the number of fatal crashes. Our findings provide valuable insights for the development of these advanced systems and their potential impact on road safety.

Publications

 
 

Additional Information

© 2026 DFG Disclaimer / Copyright / Privacy Policy

Textvergrößerung und Kontrastanpassung