A driving simulation study on driver, pedestrian and cyclist interactions in urban environment

As part of the ARCADE project, this study is aimed at identifying the main collision patterns involving VRUs in urban areas and to find the contributing factors that lead to risky conditions, also considering multiple interactions between road users. The corridor of Corso Vittorio Emanuele II in Turin, Italy, was selected for this study. Based on the official ISTAT crash database, collision diagrams were developed to evaluate the dynamic of collision between vehicles and VRUs, and then used to design the simulation scenarios. The patterns showed that the most hazardous conditions for VRU occurred on the two service roads of the corridor, and that the differences in such road layouts have led to different crash frequencies and severities. In the corridor section between Piazza Rivoli and Piazza Adriano, a dedicated cycle lane separated from ordinary traffic lanes has led to lower collisions frequencies and severities, while the corridor from Piazza Adriano to Porta Nuova, the service roads is a shared lane for both vehicles and bicycles with higher crash frequency and severities. Thirty-two participants (sixteen women) aged between 25 and 65 years, drove in different scenarios that were developed combining (i) pedestrian behaviour, (ii) pedestrian time gap acceptance, (iii) interaction type, (iv) crossing visibility, (v) pedestrian crossing start side, and (vi) lighting conditions. In particular, the whole corridor was divided into two sections. In Section 1, drivers interacted with pedestrians crossing the road at a distance from pedestrian crossings in accordance with the National Highway Code. In this section, (i) pedestrian behaviour (yielding priority vs. non-yielding priority), (ii) pedestrian time gap acceptance (3 and 6 s), and (iii) lighting conditions (day vs. night) were tested. In Section 2, four two-level factors were considered: (i) interaction type, (ii) crossing visibility (crosswalk vs. far from crosswalk), (iii) lighting condition (day vs. night), and (iv) pedestrian crossing start side (left vs. right). The driver simply interacted with pedestrians or cyclists, or in complex simultaneous interactions with pedestrians and cyclists. Several surrogate safety measures (SSMs), i.e. proportion to stopping distance (PSD), post-encroachment time (PET) and minimum time to collision (MTTC), were measured to observe participants' responses during the pedestrian interactions. Overtaking speed and minimum car-bike distance were measured in the driver-cyclist interactions. Linear mixed-effects and logistic regression models were used to analyse the results and investigate the effects of the experimental factors. Drivers behave more cautiously at night when interacting with both pedestrians and cyclists. Drivers' propensity to yield priority was dependent on pedestrian behaviour, highlighting a greater likelihood of non-yielding with prudent pedestrians. Safer interactions were accounted for when pedestrians used zebra crossings correctly and with cyclists in the shared lane, with drivers reducing their speed when close to VRUs. Driver age was the only demographic factor found significant, with middle-aged drivers being more cautious than young and older drivers, due to a balance between driving experience and perception-reaction performance. The study highlights the need to adopt safety measures to improve interaction between road users, to calm the behaviour of aggressive users and ensure better protection for the most vulnerable.