Object detection with background occlusion modeling by using multiple laser range scanners

In recent years, tragic accidents of the elderly people in a wheelchair at the railway crossing attract more public attention. However, unfortunately pedestrians and bicycles tend to get involved in the accidents, because current obstacledetectors recognize only large objects like cars and trucks. Given such situation, development of an effective method for monitoring traffic objects is highly required. In addition, if the differences between large objects (like cars and trucks) and small objects (like pedestrians and bicycles) can be recognized, railway companies have the potential to avoid unnecessary emergency stops by just making announcement or warning of security assurance for pedestrians. The prime aim of this study is to develop a novel method for recognizing moving objects in railway crossing such as pedestrians, bicycles, and cars in order to prevent traffic accidents. We apply laser range scanners with high scanning ratio, wide viewing angle and long-range measurement capability to the traffic object monitoring. Since multiple laser beams hit an identical object because of high angle-resolution, laser points are clustered based on the relative distances and specific shapes of moving objects. Although laser scanner cannot detect black car so well because of its low reflectance ratio, we tried to resolve such problem by estimating the existence of invisible objects from disappeared background data.