There is a demand by the highway safety community for better quality crash data to meet a wide variety of needs. The goal of the Federal Highway Administration (FHWA) Expert Systems for Crash Data Collection Program was to use expert systems technology to improve the accuracy and consistency of police-reported data. The program included the development and evaluation of three expert systems: (1) Seat Belt Use Derivation; (2) Vehicle Damage Rating, including Extent of Deformation; and (3) Roadside Barrier Problem Identification. Police officers used pen-based computers, containing the expert systems, to collect on-scene crash data. Embedded in the expert systems are data collection knowledge derived from experts in crash data collection and analysis. The expert systems use this knowledge to intelligently select the data to collect and assign values to elements. This knowledge is also included in on-line help screens that aid the officer in accurately identifying the physical characteristics of the crash scene. The expert systems were evaluated during two field tests. The field test results showed that the expert systems: (1) were well accepted by the officers, (2) were validated by experts in the expert system domain areas, and (3) the officers collected data on an average of approximately 2 minutes per expert system. (Author/publisher)
Abstract