Investigating driver’s behaviour on approaching a junction at the end of green time

Abstract

The DFT statistics (2010) revealed that red light running behaviour causes approximately 10 casualties per day in the UK, which is 3.4% higher than the previous year (2009). There has been uncertainty about the reasons for these violations; past literature has suggested that it could be due to insufficient amber duration or factors associated with the road environment, drivers and driving behaviour. Despite the underlying causes of these violations, red light running behaviour has been found to be more prevalent amongst younger drivers. A driving decision behaviour framework that captures continuously driver speed and acceleration performance, and their decisions was proposed. This research identifies contextual variables that can be used to predict driver’s decisions at junctions during the amber onset (such as presence of pedestrians and heavy vehicles). In particular, drivers were more likely to cross the junction when there was a vehicle directly ahead of them. Studies of drivers were conducted in a STISIM driving simulator. A methodology was developed to categorise driver responses as safe or unsafe, and to systematically assess the performances of different interventions. The advanced signal intervention (with a set of advanced signals placed upstream on the same approach of the original traffic light displaying the impending signal status from the next second) was the most effective intervention to increase stopping decision (with maximum deceleration rate < 4.9m/s2) without elevating driver uncertainty at junctions. The flashing amber intervention (with a standalone flashing amber light adjacent to the original traffic light activated 1s prior to the amber onset) however seemed to encourage drivers to stop early. Drivers braked significantly earlier when compared to other interventions and the control condition (i.e. baseline scenario). The extended amber intervention (with 4s amber phase) has slightly reduced unsafe stopping behaviour when compared to the control condition, but the intervention also increased driver uncertainty. 70% among the stopping decisions were categorised as unsafe (with maximum deceleration rate > 4.9m/s2), and therefore may not be recommended. The positive effects of the interventions to reduce violations were negated at urban junctions, which suggest that red light countermeasures may not be required at urban junctions. The findings also revealed a slower braking response time to the interventions when the drivers were within close proximity to another vehicle, thus highlighting the contextual effects of their preceding vehicle as suggested from the observational study. Future research should be extended to assess the performance of the advanced signal intervention to different levels of traffic flow and turning manoeuvres. Larger sample of drivers should be employed for improved reliability.

More information

Identifiers

ORCID for Neville A. Stanton: orcid.org/0000-0002-8562-3279

Catalogue record

Export record