Magdalena Lindman

Evaluation of the crash mitigation effect of low-speed automated emergency braking systems based on insurance claims data

ABSTRACT Objective: The aim of the present study was to evaluate the crash mitigation performance of low-speed automated emergency braking collision avoidance technologies by examining crash rates, car damage, and personal injuries. Method: Insurance claims data were used to identify rear-end frontal collisions, the specific situations where the low-speed automated emergency braking system intervenes. We compared cars of the same model (Volvo V70) with and without the low-speed automated emergency braking system (AEB and no AEB, respectively). Distributions of spare parts required for car repair were analyzed to identify car damage, and crash severity was estimated by comparing the results with laboratory crash tests. Repair costs and occupant injuries were investigated for both the striking and the struck vehicle. Results: Rear-end frontal collisions were reduced by 27% for cars with low-speed AEB compared to cars without the system. Those of low severity were reduced by 37%, though more severe crashes were not reduced. Accordingly, the number of injured occupants in vehicles struck by low-speed AEB cars was reduced in low-severity crashes. In offset crash configurations, the system was found to be less effective. Conclusions: This study adds important information about the safety performance of collision avoidance technologies, beyond the number of crashes avoided. By combining insurance claims data and information from spare parts used, the study demonstrates a mitigating effect of low-speed AEB in real-world traffic.

Using Random Forests for Data Mining and Drowsy Driver Classification Using FOT Data

Data mining techniques based on Random forests are explored to gain knowledge about data in a Field Operational Test (FOT) database. We compare the performance of a Random forest, a Support Vector Machine and a Neural network used to separate drowsy from alert drivers. 25 variables from the FOT data was utilized to train the models. It is experimentally shown that the Random forest outperforms the other methods while separating drowsy from alert drivers. It is also shown how the Random forest can be used for variable selection to find a subset of the variables that improves the classification accuracy. Furthermore it is shown that the data proximity matrix estimated from the Random forest trained using these variables can be used to improve both classification accuracy, outlier detection and data visualization.

An evaluation of the real-world safety effect of a lane change driver support system and characteristics of lane change crashes based on insurance claims data

ABSTRACT Objective: Lane changes, which frequently occur when vehicles travel on major roads, may contribute to critical situations that significantly affect the traffic flow and traffic safety. Thus, knowledge of lane change situations is important for infrastructure improvements as well as for driver support systems and automated driving development projects. The objectives of this study were to evaluate the crash avoidance performance of a lane change driver support system, the Blind Spot Information System (BLIS) in Volvo car models, and to describe the characteristics of lane change crashes by analyzing detailed information from insurance claim reports. Methods: An overall evaluation of the safety effect of BLIS was performed by analyzing crash rate differences in lane change situations for cars with and without the optionally mounted BLIS system based on a population of 380,000 insured vehicle years. Further, crashes in which the repair cost of the host vehicle exceeded approximately US

Definition of run-off-road crash clusters - for safety benefit estimation and driver assistance development

1,250 were selected and compared. Finally, the study examined different precrash factors and crash configurations, using in-depth insurance claims data from representative lane change crash cases including all severity levels in a population of more than 200,000 insured vehicle years. Results: The technology did not significantly reduce the overall number of crashes when all types of lane change crashes and severity levels were considered, though a significant crash-reducing effect of 31% for BLIS cars was found when more severe crashes with a repair cost exceeding US

Collision Warning with Auto Brake - A Real-Life Safety Perspective

1,250 were analysed. Cars with the BLIS technology also have a 30% lower claim cost on average for reported lane change crashes, indicating reduced crash severity. When stratifying the data into specific situations, by collecting precrash information in a case-by-case study, the influence of BLIS was indicated to differ for the evaluated situations, although no significant results were found. For example, during general lane change maneuvers (i.e., not while exiting or entering highways or during weaving/merging situations) the crash rate was reduced by 14%, whereas in weaving/merging situations the crash rate increased. Conclusions: The insurance data analyzed provided useful information about real-world lane change crash characteristics by covering collisions in all crash severities and thus revealed information beyond what is available in, for example, data sets of police-reported crashes. This will guide further development of driver support systems. For crashes with repair cost exceeding US

The Effect of a Low-Speed Automatic Brake System Estimated From Real Life Data

1,250, a significant crash reduction was found, although the technology did not significantly reduce the total number of lane change crashes. An average lower insurance claim cost for cars equipped with the BLIS technology also indicated that the technology contributes to reduced crash severity even if crashes were not totally avoided. Stratifying the data into different lane change crash situations gave indications of the condition-specific performance of the system, even if the results were not statistically significant at the 95% level.

A Method for Estimating the Benefit of Autonomous Braking Systems Using Traffic Accident Data

Single-vehicle run-off-road crashes are a major traffic safety concern, as they are associated with a high proportion of fatal outcomes. In addressing run-off-road crashes, the development and evaluation of advanced driver assistance systems requires test scenarios that are representative of the variability found in real-world crashes. We apply hierarchical agglomerative cluster analysis to define similarities in a set of crash data variables, these clusters can then be used as the basis in test scenario development. Out of 13 clusters, nine test scenarios are derived, corresponding to crashes characterised by: drivers drifting off the road in daytime and night-time, high speed departures, high-angle departures on narrow roads, highways, snowy roads, loss-of-control on wet roadways, sharp curves, and high speeds on roads with severe road surface conditions. In addition, each cluster was analysed with respect to crash variables related to the crash cause and reason for the unintended lane departure. The study shows that cluster analysis of representative data provides a statistically based method to identify relevant properties for run-off-road test scenarios. This was done to support development of vehicle-based run-off-road countermeasures and driver behaviour models used in virtual testing. Future studies should use driver behaviour from naturalistic driving data to further define how test-scenarios and behavioural causation mechanisms should be included.