Converting traffic signals to 100% digital detection

Colorado Springs reduces traffic congestion

David S. Zelenok
Group Support Manager
City of Colorado Springs, Colorado

Traffic congestion is one of the most controversial topics in the public works profession today. Clearly, there are many needs and issues contributing to the problem-urban sprawl, funding shortages, environmental clearances, public input processes, subdivision planning...the list goes on and on.

The problem
Sadly, most of the major highway improvements now under design will take years or decades to complete. Worse yet, the delays imposed by the public approval process and seemingly "ancillary issues" can inflate these projects' costs in turn, resulting in even more delays. In one recently released survey, the Colorado Springs area ranked highest in the nation in traffic congestion for cities of 500,000 and smaller. Although an April ballot item proposing a tax increase for capital improvements narrowly failed at the polls, traffic congestion and growth remains our citizens' number one concern.

While far from a panacea for solving all of the City of Colorado Springs' (population 440,000) traffic problems, improving traffic signal coordination has taken "center stage" in many public discussions. In response, the City of Colorado Springs has undertaken a new strategic initiative to modernize its traffic signal system by using a "cutting edge" system now gaining increasing recognition with agencies seeking to maximize the efficiency of their transportation infrastructure without resorting to time-consuming and costly projects. This article summarizes the efforts now underway to improve traffic flow and mitigate congestion on Colorado Springs' roadways, with a particular focus on its state-of-the-art "Machine Vision Processing" (MVP) technology.

Background
City staff has been working for a number of years to improve signal timing and coordination. Major efforts have been placed into interconnecting nearly all of the City's 480 signalized intersections with a central master computer with the ability to remotely program and monitor intersection operations. To better assist drivers in trip routing and decision-making during periods of congestion, a new Variable Message Signage system is under construction on the major north-south arterial (Interstate 25), along with a trail blazer system of programmable signs on detour routes. By combining these improvements, it is hoped that not only can drivers be forewarned that there is "congestion/accident ahead," but may be given suggested detour routes onto nearby major arterial streets. Once there, the interconnection benefits take over by automatically re-coordinating the signals on the detour route to favor, for example, northbound progression to replace capacity lost on the interstate's northbound lanes. The City has already implemented live video of the major highways on a local cable TV channel, and the City website displays still images updated every few minutes.

In response to citizen concerns, the City Council authorized a number of additional city staff positions in the 2001 budget, dedicated to re-timing signals, especially during the lower traffic volume times. The existing staff will continue to optimize signal timing during peak hours where much work has already been done to maximize flow. The new staff will focus their attention mainly on reducing delays on side streets before or after the peak travel periods when most of the complaints are received. Last year alone, the City implemented 50 new peak-hour signal timing plans on arterial roadways, producing an average of 15 percent improvement in travel times on those corridors. Despite these improvements, complaints have continued and staff has been searching for even more ways of improving traffic flow-without adding new highways.

Emerging technologies
Until recently, state-of-the-art vehicle detection involved placing a wire sensor into the pavement. These wire induction loops sense the presence of a vehicle on the pavement and send this analog information to the intersection controller, which, in turn, decides which "leg" of an intersection should receive a green or red signal indication.

Colorado Springs' winter season (with some of the nation's highest number of annual pavement freeze-thaw cycles combined with repeated applications of anti-skid abrasive materials to improve wintertime traction), frequent utility cuts at intersections, and increasing traffic congestion have made it very difficult to keep all 3,500 sensors operating reliably. As a general rule, five to ten percent of the in-pavement sensors have been in need of replacement each year. They can take anywhere from a few days to replace in the summer to as long as six months during winter months, as low temperatures often dictate when the work can be done.

It was concluded that the best signal-timing plans mathematically possible could be implemented, but the problem itself-traffic congestion-could not be improved significantly until the entire equipment network could be made more reliable. Surprisingly, a single malfunctioning loop detector can impair traffic progression miles away, and for weeks or even months at a time. Until recently, very little could be done to mitigate this problem, as reasonably-priced options were simply unavailable or did not meet the reliability needs of the system.

In the last year, a new technology has emerged that replaces the in-pavement wire sensor. It uses a black and white camera and computer software called "machine vision processing" that recognizes vehicles as they approach intersections. A major advantage of this new system is that the detector is not buried in the street, but rather, mounted on the traffic signal poles. This means that a "machine vision" detector can be repaired in hours, not months, with the added advantage of our crews spending less time on the street blocking traffic and resulting in even less congestion impacting the public.

Each detection loop can be replaced with dozens of computer-generated detection zones at every intersection approach. With four cameras at a typical intersection, more than 100 zones can be electronically designated at each location. More importantly, these zones may be varied, not only in number, size, and location, but based upon time of day, traffic accidents, construction zones, major events, traffic volumes, and similar considerations all without leaving the office.

Because the system uses a relatively low resolution camera, digital real-time signals can be sent on demand to the City's traffic operations center, allowing video monitoring of any signal in the entire network-2,000 video detectors. With the new system in place, traffic technicians can prepare a new timing plan, install it from their office, and then immediately observe the impact on traffic flow. By varying the number, size, and location of the detection zones, previously impossible features can be installed, such as turning off a left turn arrow based upon a low number of detected vehicles in each cycle.

Zones can be laid out in any number of patterns. The sensors also can give accurate traffic counts and even indicate the speed and assist by providing other information such as turning movements and safety-related data. Other possibilities to be studied by the Traffic Engineering staff include detecting pedestrians at school signals where the students neglect to push the pedestrian button, and using the video detection system to extend the all-red interval if it detects a vehicle likely to run a red light.

Lessons learned
The amount of travel time savings or economic benefits gained on a particular corridor will be difficult to quantify because the technicians will never be able to conclusively predict which loop would have gone out of service. Nonetheless, the value of a fully functioning and reliable system is certain to have a significant impact on mitigating congestion and improving corridor progression. Clearly, the economic benefits to the community will outweigh the costs.

Maintenance costs and operational issues will be unknown until the system has been installed for quite some time. The only citizen complaint received to date has been a single individual whose backyard is now on a direct line with the field of view of a camera sensor. The individual was invited to see what the sensor can and cannot see first-hand and, after explaining the detectors cannot zoom, pan, or tilt, the issue was resolved. Other concerns resolved to date include detection during periods of fog, low light, or heavy precipitation. Typical street light designs provide enough illumination for nighttime detection, and electric heaters are included in the camera objective lens assembly to melt and snow or ice buildup that might interfere with the detectors' visibility.

The City and federal governments have programmed $6 million toward this project. A contract was awarded in March 2001 for the complete system replacement and technical staff training. As of this report the City is anticipating delivery of the first operational group of cameras and hardware in early summer. In the meantime, intersection surveys and signal modifications are already underway where necessary. The first priority is to install the system at those intersections that presently have defective in-ground detection systems. Following those locations will be high-volume corridors, and all newly-installed signals will receive the MVP system. It is expected that the contractors will be able to install about six locations per week and complete the total project in about two years.

Summary
At this point the possibilities now seem limitless, and Colorado Springs is one of the first major cities in the country to completely convert its analog-based system in favor of the new digital one. This innovative approach to managing our transportation network illustrates the City's commitment to maximizing the efficiency of the entire 1500-mile transportation network.

With the additional staff and the flexibility of video detection, the traveling public should see a significant reduction in delays for years to come.

For more information, please contact David Zelenok at 719-385-5907 or at dzelenok@ci.colospgs.co.us.