Driving data from multi-human-in-the-loop simulation experiments
Wu, Guoyuan (2022), Driving data from multi-human-in-the-loop simulation experiments, Dryad, Dataset, https://doi.org/10.6086/D18X0V
Freeway ramp merging involves conflict of vehicle movements that may lead to traffic bottlenecks or accidents. Thanks to advances in connected and automated vehicle (CAV) technology, a number of efficient ramp merging strategies have been developed. However, most of the existing CAV-based ramp merging strategies assume that all the vehicles are CAVs or do not differentiate vehicle type (i.e., passenger cars vs. heavy-duty trucks). In this study, we propose a decentralized cooperative ramp merging application for connected vehicles (both connected trucks and connected cars) in a mixed-traffic environment. In addition, we develop a multi-human-in-the-loop (MHuiL) simulation platform that integrates SUMO traffic simulator with two game engine-based driving simulators, allowing us to investigate the interactions between two human drivers under various traffic scenarios. The case study shows that the decentralized cooperative ramp merging application, which provides speed guidance to the connected vehicles involved in ramp merging, helps increase the time headways of the involved vehicles and smooths their speed profiles. With the speed guidance, the median minimum time headway for the yielding car on the mainline increases by 57%. Also, its speed variation decreases by 17% while the speed variation of the merging truck from the on-ramp decreases by 19%. These results demonstrate the potential for the proposed application to improve the safety and efficiency of ramp merging for heavy-duty trucks, which will be particularly useful at on-ramps with relatively short merging lanes. The experiments conducted also validate the effectiveness of the developed MHuiL platform for human factor research.
The dataset was collected from the custom-built multi-human-in-the-loop simulation platform by the research team. This platform consists of a microscopic traffic simulator, SUMO, and two game engine-based driving simulators (one for trucks and the other for passenger cars) using Unity. The data has been processed with Python code to capture detailed driving information and surrounding vehicles (with respect to the two driving simulators) information.
The trajectories of each driver and the surrounding vehicles are saved in a txt file using JSON format, which should be easily accessible by most programs/software.
U.S. Department of Transportation