Connected Vehicles- Infrastructure UTC
چکیده
The goal of the Infrastructure Safety Assessment in a Connected Vehicle Environment project was to develop a method to identify infrastructure safety “hot spots” using connected vehicle data. The premise was to show that using connected vehicle basic safety messages to detect hot spots could allow for a quicker detection than traditional methods using police-reported crashes. This is because the basic safety message may be able to detect events that police normally cannot obtain, including unreported crashes and near-crashes. The project successfully developed a model to detect crashes/near-crashes, and also designed a methodology to apply hot-spot identification. Based on the work described in this report, the UTC is fully prepared to apply the methodology to data collected on the field test bed. Background Transportation agencies devote significant resources to analyzing crash data collected by responding police agencies to identify “hot spots” – locations that experience larger than normal numbers of crashes. An example of a hot spots map created based on crash data from Ann Arbor Michigan is presented in Figure 1. In many cases, upon identification of a hot spot, field investigation will point to a feature of the infrastructure that is contributing to the crashes. This feature may then be addressed specifically to improve safety. This method, detailed in the Highway Safety Manual’s (HSM) Roadway Safety Management Process (Figure 2), has been used for many years, and has proven to be effective (1). However, this method also has significant shortcomings. One of the key shortcomings is that the agency must wait for a large number of crashes to accumulate before a hot spot may be identified. In other words, this is a very reactive method that requires a number of crashes to occur before corrective action may be taken. Figure 1. Mapping of Crashes With Potential Hot Spots in Ann Arbor. Figure 2. Roadway Safety Management Process, (1) Fortunately, there is a reason that crashes are most often referred to as “accidents.” They are infrequent, even at most hot spot locations. Thus, for a statistically significant accumulation of crashes to occur requires a rather long period of time. Furthermore, accurate capture of the location of crashes has long been a challenge in the transportation community. Police reports have been notoriously inaccurate in terms of crash location – although this is improving somewhat with the use of GPS. Thus, even when a hot spot is identified, the exact location of the problem is often difficult to pinpoint. Thus, there is a need to develop a more proactive way to accurately identify “hot spots” – locations that require modifications to the transportation infrastructure to improve safety. The premise behind this project is that for every actual crash, there also exist numerous “near misses” where drivers’ take last second, extreme evasive action (such as swerving or rapid deceleration) to avoid a crash. These near misses may be as significant as actual crashes in terms of indicating potential safety problems. The challenge lies in identifying and compiling these near misses (since they have never been formally reported by individuals or through the police). However, with vehicles in a connected vehicle (V2V) environment, basic vehicular operation data will be available from the vehicle data bus. If significant evasive maneuvers may be extracted from this data, along with the corresponding GPS location, this near miss data may be analyzed by a transportation agency in a manner similar to current police crash reports to identify hot spots. Using connected vehicles, instead of police reports, offer the potential for a much quicker and more accurate network screening step as shown in Figure 2, which in turn speeds up the entire Roadway Safety Management Process. This project will analyze data from past field tests to develop prototype algorithms for hot spot identification from vehicular operations data. These algorithms will then be demonstrated and tested in the CVI-UTC Northern Virginia connected vehicle test bed to determine if they successfully extract “near miss” maneuvers. This data will then be analyzed to determine if hotspots may be identified. Then, finally, these hot spots will be examined in terms of traditional crash data to determine if there is a correlation – thus pointing to the potential of this approach.