Driver perception-reaction times in level 3 automated vehicles

Abstract

In Level 3 automated vehicles the driver is allowed to engage in secondary tasks however the driver must be ready to re-engage in the driving mode if alerted to do so when roadside circumstances exceed the capacity of the automated vehicle technology. The scope of the research was to establish the Perception-Reaction Time (PRT) of drivers in a simulated Level 3 vehicle and to examine the interdependency between the person-specific characteristics in relation to different scenarios featuring different in-vehicle distractions and different type of alerts and subsequently to compare these values with those of standard specifications used in road design in different countries for the calculation of Stopping Sight Distances (SSD). Such PRT is important because the driver needs to be alerted in a timely manner, translated to sufficiently long approach distance in advance of a critical situation, to allow for the safe handover from automated to manual vehicle control when the roadside scenarios are beyond the capacity of the vehicle automation or where there is an unexpected roadside scenario such as new traffic management arrangements of road works. The data required for the scope of this research was collected through a web-based survey which included the collection of demographic information about the respondent in the first section and a driving simulation in a Level 3 automated vehicle in the second part. The PRT of the driver was taken from the moment of the alert to the moment that the participant reacted by clicking on an on-screen box. The results of this research document gave an average perception-reaction time of 4.23 seconds based on the 85th Percentile values of the datasets and showed that the younger age groups have lower PRTs for all scenarios than their older counterparts both for different alerts (and same secondary task) and for same alerts (but different secondary task). The results also showed that the multisensory alert advantage over the visual alert is effective only up to the point determined by the demand on the cognitive resources of the participant where, in this research, such point was reached when the secondary distraction was reading and typing of a text message.