Radar Detection of Vehicles in a String: Gaining Situation Awareness of a Propagating Conflict

A database of empirical driving measurements is employed to compute the situation awareness benefit of a driver-assistance function that uses the ability of radar to detect targets several deep in a string of vehicles ahead in the same lane. The database resulted from a new roadside video imaging method that has yielded the complete X-Y trajectories, state response variables, and intervehicular motion relationships of 30,500 vehicles on a segment of five-lane arterial street. These data were compiled as a relational database, and queries were run for detecting all cases in which a string of three vehicles is disturbed by a conflict that arises in the leading vehicle pair. Computations reveal, through distributions of the actual motion responses, the larger time cushion that would be available if any following driver in a string became aware of an impending headway conflict as soon as it develops ahead of the preceding vehicle. Results show the value to the following driver that might accrue from a crash-warning function that employs such anticipatory detection. The results indicate that the threat of rear-end crashes in vehicle strings could be mitigated by driver-assistance devices that sense range and range-rate relationships several deep in the string. Because most modern radar technologies have some degree of this capability but do not use it, the results encourage the development of crash-warning applications based on the look-ahead feature.