Effects of the cross-section on the driver's behaviour approaching bicycle crossroads

This paper reports on the results of field surveys conducted at the approaching section of two bicycle crossroads in urban environment. The crossing roads are characterised by different cross-sections (one-lane and two-lane configurations) having a same marking scheme. The aim of the surveys was to evaluate the effectiveness of the specific marking scheme for the two different cross-sections and to provide relevant insight to improve the cyclist safety, based on the observed drivers’ behaviours. More specifically, the data analysis was aimed at: (i) studying the yielding behaviour of the drivers on the two bicycle configurations and evaluating the effects of the observed vehicle dynamics constraints in approaching phases; (ii) analysing drivers’ braking during the interaction with cyclist; (iii) defining a logistic regression model for predicting the drivers’ likelihood of yielding as a function of the variables describing his behaviour and the cross-section configurations. Results have stressed out a more critical driver yielding behaviour and braking performance when approaching the bicycle crossroad on the two-lane configuration. The logistic regression model have pointed out the cross-section variable to have the strongest effect on the logit of drivers yielding, determining a significant reduction of the likelihood of yielding if the interaction driver-cyclist occurs in the two-lane configuration. Overall, the obtained outcomes highlight a higher risk exposure of the cyclist crossing the two-lane configuration. The obtained outcomes imply that the same marking scheme on the two different cross-section configurations is inappropriate. It fails to induce proper driver behaviour when he/she approaches the bicycle crossing on the two-lane configuration in order to limit the cyclist's risk exposure to the values observed on the one-lane configuration. Eventually, several measures and recommendations have been defined to reduce the risk conditions identified in two-lane configuration.