Background and Motivation: Transport of persons and goods is a major source for Green House Gases (GHG) emissions and contributes to global warming. Globalisation has increased the worldwide demand for transport and solutions to achieve global GHG reduction goals are being intensively developed. In addition, urbanisation concentrates the GHG emission problem on cities, therefore solutions for urban areas provide large potentials for a green future. Highly automated driving will be a key element for smart mobility in smart cities, enabling for totally new concepts of transport. Such solutions include multi-modal and intermodal transport systems, robot taxi services and goods deliveries. However, a main obstacle for market introduction of this disruptive technology is safety validation, proofing its superiority in vehicle control compared to human drivers. Today, the consensus of the community working on safety validation is to shift major parts from on-road testing to X-in-the loop test methods and cost and time efficient virtual testing. GOALS and INNOVATION: TRIDENT develops a virtual test framework where complex but also realistic test scenarios for safety validation are created by microscopic traffic flow simulation (TFS). However, a major drawback of this method is the calibration of the driver models controlling the vehicles in the TFS. Whereas knowledge in longitudinal (speed) and simple lateral control (lane keeping) is available, modelling of human driver behaviour in complex multilane intersections is widely unknown. Deep learning will be applied to analyse empirical vehicle trajectories for calibration of driver models in TFS. In addition, sociocultural differences require different calibration of the driver models in the TFS: Whereas AUSTRIA represents typical human driving behaviour in Western culture, the ZHEJIANG province features human driver behaviour typical for Asian mega-cities. Including those socio-cultural aspects in modelling complex human driving behaviour represents the major innovation. RESULTS: TRIDENT develops a method to calibrate human driving behaviour using the use-case of lane changes in exit and entries on motorways and major urban roads, especially with multilane configurations. The virtual test framework in Austria will be based on co-simulation between the software packages IPG-CarMaker and PTV Vissim, validated on two selected traffic sites in Graz, Austria. The same is done for the Zhejiang province using GaiA software from PilotD Automotive and two traffic sites in the province of Zhejiang.
|Effective start/end date||1/01/22 → 31/12/24|
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.