Scene complexity and lighting effect driver error during train simulation
This study examined the effects of scene lighting and scene complexity (i.e. signal density) on driver performance in a train simulator (Auran simulation software running on a PC connected to a large screen data projector). 19 university students were individually trained on this simulator over the first hour. The experiment was then conducted over the following two hours and consisted of four half-hour trips (presented in a different random order to each participant): (i) simple scene with day lighting; (ii) complex scene with day lighting; (iii) simple scene with night lighting; and (iv) complex scene with night lighting. Participants were instructed to maintain as high an average speed as they could without exceeding the posted speed limits or passing through red signals. Participant ratings of alertness and interest were obtained every 5 minutes and a time-stamped log of all driving errors was also obtained for each trip. As expected, reported alertness and interest were found to decrease significantly as the time from the start of the simulation increased. The likelihood of driver error was also found to increase significantly as the trip duration increased and then plateau after 10 minutes had elapsed. While scene complexity and scene lighting were found to have little effect on driver alertness/interest ratings, an interesting interaction between complexity and lighting was observed on driving errors. Specifically, errors were more likely to occur during complex scenes with day lighting than under any other combination of lighting and complexity. We interpret this finding as indicating that participants were more easily distracted during complex scenes with day lighting because more scenery was visible.