Cognitive State Modeling of Human Machine Interactions in a Level III Driving Simulator

The development of safer and more efficient driving systems has gained attention in the field of automation and importantly, human-automation interaction. It is widely held that autonomous vehicles with embedded models of human cognition may be better able to interact with human passengers or drivers and improve overall safety and comfort. Therefore, this research aims to understand how humans interact with autonomous vehicles, with particular emphasis on modeling how humans’ trust, mental workload, risk perception, and self-confidence change in response to different stimuli. The goal is to develop an SAE Level III autonomous vehicle simulator which will serve as a testbed for human subjects experiments. The backbone of the high-fidelity 3D simulator is developed using Unreal Engine 5 and encompasses an expansive open world map set in a dynamic city environment. The simulator includes intricately designed traffic systems that mimic real-life scenarios,giving experimenters complete control over independent variables such as automation transparency, task complexity, recommended control mode, and system reliability. Various optimization techniques are applied to ensure optimal performance of the software and hardware. Pilot studies allow for the evaluation of the ecological validity of the developed scenarios, and help in identifying bugs. Future work involves data collection through experimental trials and using the collected data for gray-box system identification and validation. Citation: Wang, X., Jeevanandam, S.,Williamson, M., & Jain, N. (2023, July 27). Cognitive State Modeling of Human Machine Interactions in a Level II Driving Simulator [Oral Presentation]. 2023 Summer Undergraduate Research Symposium,West Lafayette, IN, United States.