Center for Data Reduction and Analysis Support
The Center for Data Reduction and Analysis Support (CDRAS) supports standardized access to and analysis of numerous naturalistic driving study data sets (currently 2.5 PB) housed at VTTI for researchers within and outside of the institute. Services include coding of video and audio data, data quality assurance, data standardization, data mining, event selection, and data analysis. The center actively supports data analysis collaborations with external institutions.
SHRP2 NDS Data Dissemination and User Technical Support
Projects are focused on the accessibility and application assistance of general use data files and interfaces to support data users, including those that are not SHRP 2 contractors. As part of this effort, reduced or de-identified files such as trip files, event files, and other reduced or aggregated files are developed to provide easier access to the SHRP 2 naturalistic driving study data. Completed projects have included the production and distribution of early results from the NDS data, data documentation, webinars, workshops, and other training and communication activities. Read More
Support for FHWA Safety Training and Analysis Center
The SHRP2 data represent an extensive collection of detailed information describing drivers, vehicles, trips, and roadways. The driver, vehicle, and trip information is referred to as NDS data, while the roadway data are referred to as Roadway Information Database (RID) data. Read More
Generation of De-identified SHRP 2 Trip Origin and Destination Locations
The objective of this project was to describe the nature or purpose of a trip without revealing personally identifying information. A method to retrieve generic location categories from the Google Places API was developed and applied to the entire SHRP2 data set. Read More
Linking Historical Weather Data with SHRP 2 Trip Data
The objective of this project was to accurately link historical weather data with SHRP 2 trip data. To facilitate this linkage, a searchable data structure was developed to assign weather attributes to segments of SHRP 2 trips. The grid structure, seeking to balance spatial and temporal resolution, consisted of a spatial resolution ~1 km combined with a temporal resolution of 1 minute. Read More
Developing a SHRP 2 Data Set for Machine Learning Model Training
The application of machine learning and computer-vision technologies in the automotive domain continues to garner significant interest. The SHRP 2 Naturalistic Driving Study (NDS) is a useful data source for the development of machine learning models that are performant and robust to scenarios encountered in real-world driving. To this end, a collection of driving scenarios of interest to machine learning developers was compiled from the SHRP 2 NDS, documented, and made available for distribution under a data use license. Read More
Green Low-Speed Mobility in an Aging Society
VTTI researchers were funded by the National Science Foundation to establish a group of local stakeholders to foment research on the use of advanced driver assistance systems (including automation) in order to prolong the time before senior drivers are required to cease driving based on diminished physical and cognitive capabilities. The stakeholder group will be assisting the research team in a data gathering effort to understand the potential impact of this application, identify and sidestep potential roadblocks, and establish a framework for a future deployment. The project features collaborators from Nagoya University in Japan.
Can I Drive Myself Home? Driving after Wide Awake Hand Surgery
When Can Patients Safely Drive After Shoulder Surgery?
For these studies, VTTI researchers collaborated with colleagues at the Carilion Institute for Orthopaedics and Neurosciences. Researchers first assessed the impact of Carpal Tunnel Decompression Surgery on driving performance. The results of the study, published in The Journal of Bone & Joint Surgery, will allow physicians to make evidence-based recommendations to their patients in terms of their resumption of post-surgery driving – in some cases allowing that resumption right after surgery. Read More
Data Reduction Group
The Data Reduction Group serves a variety of research groups who require video and/or audio analysis in the areas of driver performance and behavior metrics, scenario-based analyses, environmental characteristics, and validation support. The group’s goal is to provide efficient, accurate, and comprehensive coding of videos to meet the specific needs of each unique research project. Our work begins by conducting the necessary groundwork to prepare a video dataset for analysis and working alongside the researcher to develop a data reduction protocol. We then train and manage the required data reduction staff, conduct extensive quality control throughout the reduction phase, and prepare final sets of coded data. In addition to the group leader, the group currently encompasses an analyst, a team of coordinators and support staff, and a crew of data reductionists housed within four data reduction labs comprising over 50 workstations.
Data Analysis Support
The Data Analysis Support Group facilities analysis of naturalistic driving data by providing timely and efficient access to the datasets hosted at VTTI. The group is composed of experts in data preparation, standardization, mining, and analysis. Analysis support services are provided to any organization interested in utilizing large-scale driving datasets.
- Anne Deekens