U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
| REPORT |
| This report is an archived publication and may contain dated technical, contact, and link information |
|
 |
| Publication Number: FHWA-HRT-13-088 Date: May 2014 |
Publication Number: FHWA-HRT-13-088 Date: May 2014 |
Based on the precedent of successful NCHRP studies, the review of photographs was deemed an appropriate means of extending the current understanding of roadway departure crashes. The results of using crashworthiness data collected for different purposes were mixed and exceeded expectations. Presently, first event roadside element crashes have been reviewed, and the photographic data have been extracted. A proposed outline of the remaining datasets with their perceived impact and limitations has been compiled in the following sections.
Subsequent iterations of this study will continue to focus on FARS, NASS CDS, and NMVCCS. The census quality of FARS will provide many panoramic photographs from which to extract data but which are less relevant to the crash event. NASS CDS will provide near crash time photographic evidence predicated upon the strength of the photography and identification of salient roadside elements. Finally, NMVCCS photography has yielded the highest-quality and most crash-relevant information.
In deference to the on-scene crash photographs, the identified FARS crashes were assessed and set aside during the pilot study. During the next phase, it is envisioned that panoramic view software will be consulted to provide an image of the crash scene at a temporally discontinuous time with the crash. The benefit will be to understand roadside element placement and potential crash trajectory. The type of roadside element will be assumed to be static, with damaged elements being replaced with the same type of technology. The limitation exists in confirming the precise location of impact and forfeiting important post-crash damage condition data.
This dataset is caught in the middle of FARS and NMVCCS, with rich data but scant roadside element detail, captured weeks after the crash scene has been cleared. NASS CDS crashes after 2008 will be consulted by applying the NMVCCS filter for first event roadside element impacts. This will be the surrogate for the FARS data definition of roadway departure. As the precise crash location will have been photographed by the researcher, location identification has been assured. As this will have been near-crash time photography, many damage clues may have disappeared or the element might have been replaced by the time of investigation. Although this database may be able to provide information on roadside element type and its components, more subtle measurements and crash evidence might be lost to the meteorological vagaries of the area.
Beyond the first event roadside element impacts, many secondary events involving roadside elements exist. The rationale for excluding secondary impacts is that the most crash sensors were designed as first event technologies and are disabled upon first impact, rendering the technology inactive, as it cannot be triggered in the absence of a functioning sensor. Premature deployment might have been with good reason; however, graduated technologies deploying in stages such as airbags would have the most compatibility with reviewing subsequent roadside element events. For purposes of enlarging the dataset, additional secondary event crashes will be reviewed to determine whether they provide insight into to the roadway departure problem.
Based on the success of the pilot study, the Highway Safety Information System Laboratory may be asked to supply additional graduate student support. These students are preparing for graduate degrees in civil engineering, generally specializing in roadway transportation. This skill set makes them aware of design issues and capable of providing insights into matching the case photographs to the catalog. Even with the most accomplished engineering student, human error, limited hardware experience, and photographic clarity might affect proper codification. For this reason, the supplemental dataset produced in the pilot study will be reviewed by subject-matter experts.
As new students will be undertaking this task and learning the data extraction methodology, they will be subjected to a learning curve, similar to the one encumbering the early pilot study data acquisition. Student adaptation was not found to be an impediment to the study.
As mentioned previously, adoption of digital data capture is one goal for this project. As a result, results of the FHWA EAR Program are being monitored for information exchange.
Planned future activities include the following:
Photographic extraction was used to understand the composition and placement of roadway elements. As the feasibility study progressed, more elements were found to be extractible from the photographs. Similarly, as the pilot study progressed, extended roadway data were collected beyond the realm of safety.
The value of the data extraction pilot study was marketed to various stakeholders by means of committee briefings, working papers, and conference publications. The pilot study also sought to highlight the value of on-scene crash photography. Within this publication, the pilot study identification of candidate crashes and the filtering methodology was discussed. The crashes deemed suitable for subsequent review were classified with respect to data extraction value.
The development of the supplementary dataset began with the pilot study. Reviewing NMVCCS crash scene photographs, the variables and attributes were refined and augmented. It is understood that an on-scene data collector completely versed in roadside element design, placement, and construction would be the best data compiler. However, this would be cost prohibitive, and uniformity would vary. As a result, a photographic review model was adopted.
Summary statistics were prepared that illustrated the unexpected types of information available from photographic review and highlighted sample size issues. The extracted data are expected to benefit modelers in understanding the vehicle interactions with the roadside beyond the information acquired to date from vehicle sensors, which are generally disabled after the first off-road crash event.
In addition to the cases reviewed in the pilot study, additional subsequent event NMVCCS crashes as well as FARS and CDS crashes have been identified for subsequent review. The methodology will vary for temporally discontinuous images, but the basic identifying information should remain constant. Issues of damage will not be assessed, as damage cannot be tied conclusively to the reviewed crash. Consideration will also be given to the synthesis of data for on-scene, near crash time, and temporally discontinuous crashes. Finally, continued encouragement will be given to data compiling organizations to revive on-scene crash data collection.
