Skip to main content
Dryad logo

Analysis of intelligent vehicle technologies to improve vulnerable road users safety at signalized intersections

Citation

Xiao, Ivan Runhua; Qian, Xiaodong (2022), Analysis of intelligent vehicle technologies to improve vulnerable road users safety at signalized intersections, Dryad, Dataset, https://doi.org/10.25338/B8234N

Abstract

The project needs data for macroscopic statistical modeling, which are OTS rankings and historical crash data. 

OTS crash ranking data

California Office of Traffic Safety (OTS) provides a crash ranking dataset that was developed so that individual cities could compare their city’s traffic safety statistics to those of other cities with similar-sized populations. The OTS crash rankings are based on the Empirical Bayesian Ranking Method. It adds weights to different crash statistical categories including observed crash counts, population and daily vehicle miles traveled (DVMT). In addition, the OTS crash rankings include different types of crashes with larger percentages of total victims and areas of focus for the OTS grant program. In conjunction with the research context, two types of crash rankings are focused on, namely pedestrians and bicyclists.

SWITRS crash data

The Transportation Injury Mapping System (TIMS) to provide the project quick, easy, and free access to California crash data provided by the Statewide Integrated Traffic Records System (SWITRS). The crash data includes bicycle and pedestrian collisions with vehicles resulting in injuries from 2014 to 2018. Besides, this crash database provides detailed accident reports including information on casualties, vehicle mode, accident reason, accident location, and road condition. With this information on crashes, we will select crashes between vehicles and VRUs at signalized intersections, which is the scope of this study. To avoid misunderstanding, the crashes in the following content will only refer to accidents between vehicles and VRUs. Besides, we will also collect historical weather data (including daily temperature, wind speed, rainfall, humidity, and visibility) and road condition data. All these data will be used for the next crash feature analysis. The data is publicly available and no commitment is required from SafeTREC.

SUMO Source Code (Modified)

This repository also includes the modified SUMO source code for traffic simulation. The modification is done in two aspects. First, a series of parameters of junction-control models are added to the set of vehicle type parameters, such that the simulation scenarios for different IVTs are defined by changing the values of vehicle type parameters. Second, a filtering logic is inserted into vehicles’ interaction processes. It determines whether a potential foe object is in the blind spot areas; whether the subject vehicle’s driver is distracted in this time step; and whether the equipped IVT can compensate for the visual limitations. The section below lists all the added parameters and functions.

Methods

The Safe Transportation Research and Education Center (SafeTREC) at the University of California, Berkeley, develops the Transportation Injury Mapping System (TIMS) to provide a quick, easy and free access to California crash data provided by the Statewide Integrated Traffic Records System (SWITRS). We collect five-year-long crash data, which are from 01/01/2014 to 12/31/2018. The crash data includes bicyclist and pedestrian collisions with vehicles resulting in injuries across four types of crash severity: fatal, severe injury, visible injury, and complaint of injury. The data consists of three tables including the collision dataset, the involved parties dataset, and the victims dataset. In particular, we use the collision and parties datasets that contain enough information for modeling. The rows in the crash data are built based on each case of a crash and includes information such as weather, road surface, road condition, control device, and lighting. The parties dataset includes information specific to each vehicle or VRU such as age and sex. In order to perform a party-by-party analysis, we attach the datasets of each crash to every pair of VRU and vehicle that involved in a specific collision.

Usage Notes

The data files can be viewed by Excel. 

Funding

National Center for Sustainable Transportation Technology, Award: 65A0686 Task Order 053