A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in the Aslan Institute of Technology, Bangkok, Thailand.

Thesis (M.Eng.) - Asian Institute of Technology, 1969

Two car-following models developed by LEE (1966) have been utilized to measure the detection times of the following-car driver in a queue forming situation when brake lights were used and when they were not used. The two models are postulated on the criterion that in a line of traffic, the driver of a following car detects in the movement of the leading vehicle primarily by the physical functioning of his eyes. One model uses the time-rate of change of visual angle subtended by the leading vehicle as the stimulus:, the other uses the second derivative of the visual angle as stimulus. In both models, longitudinal acceleration of the following car was taken as the response. From the data collected, 31 for the brake lights on condition and 33 for the brake lights off condition, the relationship between the stimulus and response was tested by fitting least-squares regression lines to each run. For each run, the response data, comprising the of values, at similar intervals of time, were subjected to a series of time shifts. For every time shift, the response data were compared to the stimulus data by fitting a least-squares regression line to the observations; the coefficient of correlation was computed for each regression line. The detection time was taken as the time shift which corresponded to the highest value of correlation coefficient in a plot of the amount of time shift vs its corresponding correlation coefficient. Statistical comparisons between the correlation coefficients obtained from different models were made for different brake lights conditions. The results of the analyses show that the A(d9/dt)/At model is statistically better than the d9/dt model. Comparison of the detection time values for different brake lights conditions was made. The results show that the detection time values obtained using the A(d6/dt)/At model are significantly lower than that obtained using the d0/dt model. Using the A(d0/dt)/At model, it was found that the presence of the brake lights in the leading car significantly reduces the detection time of the following-car driver when brake lights are not used in the leading vehicle. The d6/dt model, however, gave contradictory results. It was found that the mean values of detection time are significantly similar in the brake lights on condition and in the brake lights off condition