Validation of a bicycle simulator for cyclist-vehicle interaction analysis

This study tested the validity of a bicycle simulator to analyze the behavior of cyclists, by comparing both kinematic variables and behavioral indicators observed in a real-world case study and within a laboratory experiment. The virtual environment used was a rigorous reconstruction (road geometry, signs, markings, external objects, traffic flows) of a portion of a road network located in Padua, Italy, which included a roundabout and a t-intersection, connected by a 200-meter straight road. The scenarios were defined so that participants could approach the roundabout from all directions; each subject had to approach the roundabout twice corresponding to two different levels of traffic flows (high and low). Sixty-four participants were tested at the simulator; real-world data consisted of trajectories and speeds information collected with an 8-hour video survey. Selected paths at the roundabout and crossing section, approaching speeds and accepted gaps at intersection recorded in simulated scenarios were statistically analyzed and compared with those recorded in real world. Findings from this analysis provide evidence that the driving simulator has absolute validity considering a set of diverse kinematic and behavioral variables, indicating its reliability for studying the behavior of cyclists.