DATA: Delft Automated Training and Assessment
Our mission is to
- Improve road safety by using innovative driver training and assessment methods
- Gain understanding in driver characteristics and accident causation
DATA (Delft Automated Training and Assessment) is a research group that uses behavioural observation and statistical analysis methods in computerized training and assessment. DATA is led by Prof. P.A. Wieringa and is part of the Department of BioMechanical Engineering of the faculty Mechanical, Maritime and Materials Engineering (3mE). DATA is an initiative of ir. J.R. Kuipers (Green Dino) and Prof. P.A. Wieringa.
DATA is looking for partnerships to the research activities, both within and outside the field of driver training and assessment.
Virtual Assistant (2004 – 2009)
The research of the DATA group started in 2004 with the Virtual Assistant project focusing on simulator-based training and assessment for learner drivers. The project objective was to develop a new generation car simulators, more effective than the state-of-art. In this project we recognized that simulators offer important new opportunities for driver training: free control over training conditions, standardization, and objective driver assessment. We tested participants in a driving simulator in our laboratory to investigate the effect of various forms of augmented feedback on performance and learning. Furthermore, we obtained driver performance data from driving schools. We developed a method that processes raw measurement data into meaningful indicators about the learner driver, to be used for student-adaptive feedback, instructions, and guidance. Results showed that with multivariate statistics, a reliable driver profile could be generated from the driver’s in-simulator performance. This driver profile predicted on-the-road behaviour, such as the duration until the learner driver obtained his or her driving license.
The Virtual Assistant project was funded by the Dutch Ministry of Economic Affairs, under its Innovation Program on Man-Machine Interaction, IOP MMI. Simulator developer Green Dino BV initiated the project and participated in the research. In the project, we worked with various parties, such as SWOV Institute for Road Safety Research, and the Dutch Driving Test Organization (CBR).
Simulators for road safety (2010 – 2012)
The societal costs of road accidents have been estimated at 12 billion Euro per year in the Netherlands or 2.6% of our gross national product. Young male drivers are overly involved in traffic incidents; an issue often referred to as the “young driver problem”.
Driving simulators offer new possibilities, such as objective driver assessment, standardization, and purpose-developed virtual environments. The research aims to investigate to what extent a simulator-based test is able to predict the likelihood of a car crash on the road. In addition, the project aims to develop and experimentally evaluate simulator-based training methods for improving road safety.
Data of driving simulator performance, personality and psychometric tests, and driver’s crash records, and meta-analytic results from the literature are combined in order to construct a multivariate statistical driver model. This model will indicate which types of driver characteristics are most predictive of future crash involvement. Based on the model results, a training method will be developed that focuses on suppressing deviant driving behaviours. Transfer of training experiments will be conducted to investigate whether the newly developed methodology will improve safe driving behaviour.
The research is funded by the Netherlands Organisation for Scientific Research (NWO) under a VENI grant awarded to Dr. Ir. J.C.F. de Winter. In this research we will work with - amongst others - driving simulator developer Green Dino BV, driving schools, the faculty of AeroSpace Engineering, the SWOV Institute for Road Safety Research, and the University of Valenciennes (France).
