Project

While automated vehicles are expected to have a great positive impact on traffic safety, the interaction between manual driven vehicles and automated vehicles remains unknown until their introduction in real traffic. Some of the challenging interactions occur in cut-in (manual driven car enters lane in front of automated vehicle) and rear-end approaching scenarios (manual driven car approaches automated vehicle from behind). Here, a change in occurrence for the first scenario and the discrepancy of expected behaviour (e.g., higher speed) in the second scenario can potentially lead to an increase of risk.
The data was collected within the L3Pilot1 project, where automated vehicles were tested in normal traffic (treatment) and compared with a fleet of manual driven vehicles in the same environment (baseline). Several vehicles were equipped with logging devices and driven on the city motorway around Gothenburg, Sweden. Measures of interest were indicators of surrounding vehicle driver's behaviour such as time headway (THW), distance, velocity, to determine if there are changes in their behaviour based on an automated vehicle present. We analysed 1646 cut-in and 52042 rear-end events. Both events were identified from ego vehicle perspective. Motivation for performing the manoeuvre was manually annotated for each cut-in event and distinguished between discretionary (e.g., after overtaking) and mandatory (e.g., end of lane, entry ramp).
The results show that for cut-in events the minimum THW was significantly longer in treatment than in baseline. The average speed was significantly higher in the baseline. Furthermore, cut-in events had 2.7 greater odds of being mandatory type in treatment than in baseline. Specifically, cut-ins from entry ramps occurred much more in treatment. In the rear-end events, there is a small effect on the minimum distance kept by the subsequent vehicle (slightly lower in treatment compared to baseline) and on the minimum acceleration when braking of subsequent vehicles (slightly higher for treatment which means less severe braking in average).
The lower driven speed of the automated vehicle was expected to have potential safety implications for subsequent traffic. However, the observed effects were small and did not indicate a riskier behaviour when approaching an automated vehicle compared to normal traffic. The increase in cut-in events at entry ramps indicates a system's limitation to react to entering vehicles by changing lanes or leaving space. This might lead to a decrease in acceptance.
The purpose: The aim is to stimulate project generation to maximize the benefits of SAFER multi-disciplinary platform and obtain a broad commitment, both between partners and within research areas. The data collected in the L3Pilot project are the enabler for making automated driving safe. Publishing and presenting the methodologies and the results from the project at a scientific conference and journal is a strategic competence that will help SAFER partners to continue to be leading actors within one of its core areas, Safety Performance Evaluation.
Expected results: A publication that will be presented at the International Conference on Traffic and Transport Psychology (ICTTP) on 23-25 August 2022, Gothenburg, Sweden. An extended version of the publication will be submitted to a peer-reviewed journal, for example, Special Issue to Transportation Research Part F: Traffic Psychology and Behaviour Journal or similar.