University of Maryland Department of Civil & Environmental Engineering By G.L. Chang, M.L. Franz, Y. Liu, Y. Lu & R. Tao BACKGROUND SYSTEM DESIGN DATA ANALYSIS & RESULTS PERFORMANCE EVALUATION SURVEY DATA REDUCTION Traditional dilemma zone protection systems using spot vehicle detections from loop detectors provide protection for only those vehicles traveling at the posted speed limit. Recent technology has been developed to dynamically track individual vehicles as they approach an intersection of interest. Using real-time vehicle speed and distance from the intersection, the manufacturer of this technology claim that such a system can provide dilemma zone protection for each vehicle; thus improving the overall safety of the intersection. The goal of this research was to develop a system that utilizes the dynamic detection technology and to evaluate the performance of this system at an intersection suffering from crashes susceptible to correction by sufficient dilemma zone protection. Vehicle Detection that Called the All-Red Extension Vehicle Position After 3 Seconds of Red Map of System Design Acknowledgements: Thank you to the Maryland SHA for sponsoring this research Thank you to Mr. Bob DeBlase of SHA for his input on the signal logic Thank you to Mr. Larry Gredlein of SHA for providing temporary traffic control during tube installation Thank you to our colleagues at the Traffic Safety & Operation Lab for assisting in data collection and data reduction Synchronizing Equipment with GPS Video Reduction Software Summary of Data Analysis Vehicle Position at the End of the All-Red Extension Map of Equipment Location for East Bound Data Collection RESEARCH MOTIVATION The site selected for installation of the dynamic dilemma zone protection system was the high speed (55 mph) rural intersection of US40 at Red Toad Road in Cecil County, MD. This site was selected based on the frequency and severity of crashes that may be corrected by dilemma zone protection (i.e. right angle crashes). From the years , the intersection of US40 & Red Toad Road experienced 89 crashes, 40 (45%) of which were right angle crashes. PRE-DESIGN SURVEY To properly design the dynamic dilemma zone protection system, a pre- design survey was conducted at the target intersection. The survey collected data on vehicle speeds and reaction to the yellow interval. The following table and figures summarize driver behavior during the yellow interval on the major approach (US40): Summary of Vehicle Speeds During the Yellow Interval Speed Distribution During Yellow Interval (By Direction) Observing the wide distribution of speeds during the yellow interval suggests that traditional dilemma zone protection provided by static loop detectors may not be appropriate. This realization is illustrated in the following figure, showing the dilemma zone location for various approach speeds, calibrated to observed driver behavior at this intersection: Dilemma Zone Locations for Various Speeds To address large speed variance of approaching vehicles during the yellow interval, a system state-of-the-art microwave sensor that dynamically tracks approaching vehicles was selected for implementation. The data collected from these sensors was used to control the signal logic, including green extensions, gap-outs and all-red extensions. The all red extension provided dilemma zone protection by calculating an estimated time of arrival for each vehicle during the first 3 seconds of the US40 red interval. To address the over-representation of severe crashes in the eastbound direction, a two sensor system was designed for this direction. The location and detection range of each sensor is illustrated below: System Control Logic: Call a green extension after reaching the minimum green time if a vehicle was detected within 500ft of either intersection stop bar with a minimum speed of 27 mph; Call an all-red extension if a vehicle is detected within 500ft of either US 40 approach at a minimum speed of 56 mph at the onset of the US 40 red interval. The length of the extended all-red interval is determined by the vehicle’s speed and its distance from the stop bar with a maximum extension of 2.5 seconds. Additional dilemma zone protection for EB US 40 was provided by sensor 3. The section of EB US 40 covered uniquely by the second EB sensor (from 500ft to 875ft relative to the EB stop bar) was used only for all red extension. Within this range, a vehicle must be detected with a minimum speed of 67 mph for an all-red extension to be called. Approximately 18 months after the system was installed, a performance evaluation was conducted on the eastbound direction. Data was collected for 4 hours via tube detectors, video cameras and the signal log file. The location and function for each piece of data collection equipment is presented below: The evaluation was performed by identifying missed all-red extension calls. False positives were those events in which an all-red extension was called, but no vehicles met the criteria to do so (at the cost of efficiency). False negatives were the events in which an all-red extension was not called but at least one vehicle warranted an all red extension (at the cost of safety). For the analysis of the all-red extension function, only those vehicles detected within 3 seconds of onset of US 40 red interval where considered. During the 4 hours analysis period of the eastbound direction: 521 were detected within 3 seconds of the onset of US40 red interval. Only one all red extension was called during this period This call was verified (See figures to the left) The length of the all-red extension allowed the target vehicle to clear the intersection before the side street traffic was released. The remaining 520 vehicles detected within 3 seconds of the onset of US40 did not meet the criteria for an all-red extension Other than the vehicle that called the all-red extension, no other vehicles ran the red signal Based on the results of the survey and reduction in right angle crashes since the installation of the system, the Distance From Stop BarEquipment UsedData Collected 200ftVideo Camera 2Speed, Class, Lane 275ftVideo Camera1Signal 300ftTube Detector 1Speed, Class, Lane 400ftVideo Camera 3Speed, Class, Lane 500ftTube Detector 2Speed, Class, Lane 600ftVideo Camera 4Speed, Class, Lane 700ftTube Detector 3Speed, Class, Lane 800ftVideo Camera 5Speed, Class, Lane 875ftTube Detector 4Speed, Class, Lane Summary of Equipment Used Since each piece of data collection equipment had its own internal clock, all clocks were synchronized using hand-held GPS units (Figure below, left). While the tube detectors had “out-of-the-box” software to compute vehicle speeds, the video data had to be manually reduced. Using construction cones with known spacing, software was developed to calculate vehicle speeds (Figure below, right). This software was also used to record the timing of each signal interval change.