Engaging Human-in-the-Loop for Autonomous Vehicle Simulation

Many autonomous vehicles are still in the development phase due to limited research and testing and will take a considerable amount of time to further develop before they are ready for public release. The main objective of this study is to introduce a human-in-the-loop simulation framework for supporting autonomous vehicle research. Our proposed simulation framework aims to facilitate AV assessment by providing a safer and more efficient way. Functionally, it is focused on the understanding of AVs’ operations in the presence of pedestrian users. The developed simulation framework allows a human pedestrian avatar to be integrated into the high-fidelity 3D virtual environment and interact with simulated autonomous vehicles via standard keyboard input methods or virtual reality (VR) methods. This enables safer pedestrian-AV integration research and testing, and the ability to implement a series of risky edge-case scenarios in less time and cost than would be feasible in a real-world setting.