PATTERN ORIENTED ANALYSIS OF ACTUAL DRIVING EXPERIMENTS IN DRUG EVALUATION

This paper describes a German project, whose aim was to develop a data acquisition and analysis unit, consisting of an instrumented vehicle, and an analysis system for presenting and processing video records. This development is based on a systems approach, assuming that performance under drug influence should be analysed within the interaction of the systems element driver, vehicle and traffic. The basic assumption is that combinations of different variables are nonlinear, so that their patterns may generate safety-critical situations. The test vehicle is a Mercedes Benz 300 TD, equipped with sensors and video cameras to measure driver and vehicle behaviour and the traffic environment. Data are stored in an on-board personal computer (PC). Video and data are presented together on a large high- definition monitor, and the multimedia system, consisting of a PC and video recorder, is synchronised by a time code reader. Sequences, containing patterns of typical interactions of the recorded system variables, can be given safety ratings. The sequences for the whole driving experiment can then be analysed statistically, and cross- validated with the laboratory's well-proven battery of tests. Future studies will explore the possibility of using neural nets for the pattern recognition stage. For the covering abstract see IRRD 866577.

Language

  • English

Media Info

  • Features: References;
  • Pagination: p. 658-63
  • Monograph Title: ALCOHOL, DRUGS AND TRAFFIC SAFETY-T92. CONFERENCE PROCEEDINGS

Subject/Index Terms

Filing Info

  • Accession Number: 00670011
  • Record Type: Publication
  • Source Agency: Transport Research Laboratory
  • ISBN: 3824901315
  • Files: ITRD
  • Created Date: Dec 23 1994 12:00AM