A preliminary evaluation of a speed threshold incident detection algorithm

Abstract

Traffic management centers can reduce emergency vehicle response time to incidents with immediate detection of incidents. Traffic management centers detect incidents by applying speed, occupancy, and/or volume loop detector data to an incident detection algorithm. San Antonio's new traffic management center, TransGuide, uses a new speed based algorithm, The TXDOT (Texas Department of Transportation) Speed algorithm. This research compares the TXDOT Speed algorithm with California algorithm #8 and California algorithm #8 using Fuzzy Logic to evaluate the new algorithm's effectiveness in detecting incidents on freeways. To test these algorithms, real data from TransGuide were run through the algorithms. Algorithm output were compared with CCTV (closed circuit television) recordings to determine how often the algorithms detected incidents, how long it takes them to detect incidents, and how frequently the algorithms falsely declared incidents. The results showed that the TXDOT algorithm performed best in detecting the most incidents and in producing the fewest false alarms, but it did not have the best detection times. However, the algorithms were not tested during many congestion periods. It is expected that the TXDOT Speed algorithm would produce numerous false alarms if it were tested during congestion. Recommendations were made to enhance the TXDOT Speed algorithm to improve its incident detection time and to allow it to perform well in congested areas.