| Recent national statistics indicate that in 2002 6,316,000 vehicle crashes resulted in 42,815 deaths and 2,926,000 injuries. In recent national statistics indicate that in Minnesota 94,969 crashes caused 657 deaths and 40,677 injuries; this includes 13,336 crashes on freeways alone at an estimated cost of {dollar}86,684,000 excluding death, injury, and congestion costs which are also of the same or higher order of magnitude. Traditional measures to reduce crashes include improved geometric design, congestion management strategies, as well as better driver education and enforcement. While such measures can be effective they are often not feasible or prohibitively expensive to implement. This realization along with the increasing need to reduce crashes and their side effects has recently led to proactive approaches in order to avoid their occurrence in the first place. One of the most promising options gaining wide acceptance in recent years is the concept of detecting crash prone flow conditions in real-time and warning drivers when the probability of a crash is high in order to increase their attentiveness thereby reducing the number of crashes. Research in identification and detection of crash prone traffic conditions is embryonic; for example, demonstration that such conditions actually exist is still lacking let alone a methodology for effective detection and system deployment. The research effort presented in this document aimed in detecting traffic flow conditions associated with increased crash probabilities as well as a general methodology for identifying the crash mechanism. The highest crash occurrence freeway section in the state was selected for observations and instrumented with a unique array of sensors and surveillance equipment. State of the art detection and surveillance stations were designed, assembled, and deployed generating an unparalleled database of detailed traffic measurements as well as the recording of crashes in progress. Based on this "complete" view of the problem we were able to identify real-time measurements that reliably describe crash prone traffic conditions as well as determine the causes of such conditions. The results can be used in the design and deployment of driver warning systems as well as other active measures aimed at reducing risk in high crash areas. |
