Child pedestrians have a high casualty rate, which is normally blamed on their inexperience and carelessness. The possibility that it is due to error by drivers, rather than children, has been neglected. An obvious source of driver error is that children are more difficult to see than adults because they are smaller and can be obscured more readily, as is supported by analysis of accident records. A more subtle form of perceptual error is related to the judgement of time-to-collision, the time it would take to reach a pedestrian. This could be obtained in two ways - directly from interpretation of optic flow, and indirectly from separate perceptions of distance and speed. The latter method may be of particular relevance in pedestrian accidents because time-to-collision can be much greater than that for which optic flow is an effective source. It is postulated that the resultant need to judge distance to child pedestrians leads to overestimation of distance, because drivers momentarily perceive children as larger people at a greater distance. Time-to-collision would therefore be overestimated, so the risk of collision with children would increase. This hypothesis is supported by simulation testing, and is corroborated by predicting how accident statistics would reflect this type of driver error. It is concluded that over half of all child pedestrian accidents in Britain might be avoided if the error could be remedied. Several remedial measures are proposed. The simulation tests not only support that hypothesis, but also suggest that the cognitive process by which time-to-collision is obtained from optic flow differs from that generally assumed. Instead, an alternative algorithm is proposed. The new algorithm provides an understanding of situations which were difficult to reconcile with the earlier hypothesis, such as the ability to respond to moving objects which are very small, of variable shapes, visible only briefly, or changing speed. (A)
Abstract