Analysis and assistance of driver actions in dynamic and critical driving situations during conditional automated driving

• Applicants: Professor Dr.-Ing. Steffen Müller; Professor Dr. Manfred Thüring
• Subject Area: Human Factors, Ergonomics, Human-Machine Systems ; Automation, Mechatronics, Control Systems, Intelligent Technical Systems, Robotics
• Term: from 2017 to 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 326727090

Conditional automated driving is the next level of today's driver assistance systems. Such a system enables automated driving for a limited time period and in specific situations without continuously supervising by the driver. Up to now, the Vienna Convention on Road Traffic and the ECE rules have forbidden the serial production of conditional and highly automated cars. However, this is about to change since a modification of the Vienna Convention in 2014. The new version allows conditional automated vehicles, but requires that they can be overridden or switched off by the driver at all times. Especially under highly dynamic driving conditions, interferences might lead to critical situations which would not occur if the system would be still in control. This raises the issue whether drivers interventions are always safe - and if not, what kind of assistance is needed to increase safety. Since this research issue has not yet been resolved, it is the focus in the proposed project.The first empirical study investigates drivers behavior, performance and workload when they switch from automated to manual driving during a highly dynamic maneuver. Such a maneuver requires the lateral or longitudinal control of the vehicle by the driver. The danger potential of the situation is determined by physical parameters, such as the degree of adhesion and time headway. In order to determine and vary these parameters exactly at the time of handover, drivers are explicitly prompted to take over. In contrast to this procedure, the second study aims at provoking the drivers to initiate the handover themselves. Behavior, performance and workload are captured and the physical parameters at the time of handover are measured.The results of the experiments are used to devise two different concepts of an assistance system which support drivers who want to regain control during automated driving even though driving conditions are highly dynamic and interferences might be hazardous. The basic idea for such assistance is to compensate drivers overreactions. The two versions will be tested against each other and the more promising one will be optimized on the basis of the test (if necessary).The project aims at clarifying if the demand of the revised Vienna Convention, that conditional automated systems can be overridden or switched off by the driver at any time, is adequate in all situations. To accomplish this, take-over situations are investigated in highly dynamic situations and a concept for technical assistance is proposed and tested. This concept together with the experimental results should provide valuable insights for the future development of conditional automated vehicles.Experiments in a driving simulator are limited due to the artificial environment compared to driving in a real car. Therefore, an additional study is conducted with a test vehicle, where it is investigated if the results obtained in the simulator are also valid for drivers in a real car.

• DFG Programme: Research Grants