To investigate effects of decreased visibility on cognition and reaction during rainfall while driving on an expressway, two experiments were conducted to compare drivers' cognitive and reactive characteristics in rain and non-rain conditions. For Experiment 1, participants adjusted an accelerator pedal as quickly as possible in response to acceleration or deceleration of a leading vehicle. In Experiment 2, participants momentarily perceived the traffic conditions ahead and identified the position and type of each leading vehicle. Results suggest the following: (1) Rainfall-impeded visibility decreases driver awareness of slight changes of traffic conditions and delays their risk-avoiding reactions to those changes. (2) Drivers lack attentional resources because of increased difficulty in recognizing traffic conditions ahead because of rainfall-impeded visibility. Consequently, they become more likely to commit cognitive errors. Results suggest that speed restraint during rainfall is effective not only for preventing slip accidents caused by wet road conditions, but also in compensating for delayed reaction and lack of attentional resources, which causes human errors such as cognitive errors.
Fast and precise detection of trafficincidents is required to mitigate negative effects by using traffic management measures. This study proposes an incident detection method that detects an incident from observation data of traffic detectors by using non-parametric statistics derived from historical observation data. One of the characteristics of the non-parametric statistics is that the method can work without any trial-and-error parameter calibrations. The proposed method detects rare occurrence events that seem to be caused by traffic incidents, and consists of the two components namely estimating occurrence probability of traffic state based on the non-parametric statistics and checking consistency with a traffic state caused by an incident bottleneck. The proposed model was empirically verified by using actual field data obtained from the Shibuya Line in the Tokyo Metropolitan Expressway. The verification results showed that the precision of the proposed method was comparable to the California algorithm and that the proposed model had high applicability due to simplicity of parameter calibration.