Not detecting the pedestrian at an adequate approach sight distance is one of the major contributory factors in night time crossing pedestrian involved accidents. The visibility distance is determined by pedestrian, vehicle and driver related characteristics. Pedestrian dress color, vehicle headlamp beam condition, drivers’ age, driving experience, drivers’ eye sight index, drivers’ fatigue level, and night time driving experience are considered in this research. The aim of the current research is to determine the visibility distance from pedestrian with respect to pedestrian, driver and vehicle characteristics listed above.
A circular loop road without street lamps was selected to perform the test. The on-site survey was performed using a dummy pedestrian. The test drivers were instructed to hoot the horn at the very first moment of identification and stop the vehicle on pedestrian crossing. Video recording technique was used as the method of data collection. The on board enumerator on the test vehicle recorded the variation of odometer during each. The detection distance was later calculated using the area under the curve of variation of velocity with time. Detection distances were calculated for every test driver under high and low head beam for five pedestrian clothing colors (i.e. white, black, red, yellow, and retro-reflective). According to the result, drivers’ detection distance is significantly affected by crossing pedestrian clothing color and head lamp beam condition. Under both lighting conditions, retro-reflective color is recognized longest distance by putting white color in second place. Meanwhile black color gives the lowest detection distance. Thus, wearing a white color dress is safer in night time while people who wear dark color should aware of their visibility in the night. The findings of this research will help to improve night time visibility of crossing pedestrians. Further, visibility distance estimation model was developed using the most significant influencing factors. The validity of the model was tested using Root Mean Square Error (RMSE) and the very lower value obtained justified the model validity.