Lighting for pedestrians:
Using eye tracking
To know what should be lit, we used eye tracking to find out what things pedestrians tend to look at.
Range bias is a problem with eye tracking in natural settings – the number of times a certain type of thing is looked at will depend upon the number of times that particular type of thing is encountered in the walk.
We used a dual-task approach to identify these critical fixations.
The two most important items were other people and pavement obstacles (potential trip hazards).
Analysis of the eye tracking suggests that the dual task approach avoids the stimulus range bias found when counting all fixations or when using fixation probability.
We used the eye tracking to find out how we look at other people. An initial analysis was carried out using a small sample of the original data.
We later repeated that analysis but using the whole data set.
The findings were largely confirmed: we tend to look at other people when they are approximately 15 m away and for only 0.5 seconds.
These are critical data for lighting experiments. The distance away tells us about the visual size of the target: past studies using a assumed 4 m distance or using a stop-distance procedure may have drawn incorrect conclusions. The duration tells us about task difficulty: past studies allowing unlimited observation of the target person provide an easier task and therefore an underestimate of the lighting needed.
We used the eye tracking to find out how we scan for trip hazards.
This is an interesting analysis because eye tracking identifies foveal fixation but hazard detection is peripheral vision task. Our estimate is 3.4 m ahead.
We used accident data and foot clearance data to determine the critical height of a pavement hazard – the minimum size likely to cause a trip if not seen. This was found to be 10 mm. It is smaller than the 25 mm rule-of-thumb previously used; it is therefore more difficult to see, which has an impact on the lighting needed after dark.