Traffic Light Vehicle Detection and Volume Control

Omar Ahmad, Marc Lozier, Nicholas Mons
supervised by Dr. Vadim Geurkov, Associate Professor

Traffic congestion is becoming an apparent problem in major cities, and with the rising population and lack of roadway space it will only continue to get worse. This project is aimed at targeting inner-city traffic which can easily be avoided. Drivers are spending an unnecessary amount of time at intersections when they could continue along their desired path. This lead to our group concluding that if a system were implemented correctly the wait times for drivers at intersection could be reduced, allowing them to travel along their desired path. Further problems include:

Roadwork: Cutting the roads to install the current method of traffic detection, inductive loops delays traffic at that intersection for a minimum of a week and sometimes up to a month Invasive construction: Installing inductive loops requires pavement cutting which reduces the pavement life, especially in Toronto where weather changes are rapid Traffic jams: Traffic jams caused by poorly designed algorithms and imprecise detection at intersection has caused jams all over the city. Electricity wasted: Looking at the power consumption of traffic lights, they consume approximately 5-10% of a cities power.

The design is simple. Traffic lights with cameras built in will be implemented with motion detector technology that will produce a highly accurate approximation of the number of cars turning left, going straight and making a right turn. The camera will take a picture of one part of the intersection and the image will be run into through a series of MatLab functions which will determine the number of distinct cars in the images. This will provide information on the number of cars in the left, right and center lanes. This information will be streamed to the algorithm which will determine when to switch the lights to minimize the overall city drivers wait time, stopped at an intersection.

In today's time where power production costs are high and also looking at the worsening environment other renewable power sources will be more appropriate for traffic lights. There is a need to reduce costs and power usage and therefore in this project looking at implementing other renewable power sources to save both money and energy will be presented.

Project targeted applications:

The main methods of detecting vehicles that have been implemented around the world include: inductive loops, magnetic loops, microwave radar and video image processing. Each of these technologies possesses their own pros and cons; some more cons than pros. While inductive loops are the most popular method of detecting vehicles because of their flexible design and high accuracy to count data, construction requirements and high costs to install and maintain outweigh any gains associated with this technology.

Video image processing is a more practical implementation as it less expensive and complicated and provides the necessary data to compute a proper algorithm with a high degree of confidence. If implemented properly, video image processing has the ability to: monitor multiple lanes and gather highly accurate data of all the lanes simultaneously, with sometimes as little as one camera per direction in the intersection. Installation of the video cameras will be either in the actual traffic light itself or positioned in the intersection as to receive the best image. As with any technology there are downfalls associated with video image processing. Performance may be affected if low quality cameras are used or if weather conditions are harsh enough to cause the quality of the images to decrease to a point where edge detection is no longer able to occur.

Many streetlights are equipped with cameras used strictly for viewing the intersection at the Traffic Management Center. If implemented correctly, the system would be able to detect the number of cars in each lane and compute the direction they are heading. The number of cars will run through an algorithm that will be connected to the traffic grid to decrease overall traffic times.

Solar power will be implemented as it would reduce the high power usage caused by conventional or LED traffic lights. An effort is being made where traffic lights are being changed over to use LED lights, but if at the same time solar panels are installed in the system, the operating costs will reduce to zero. The only costs involved then would be the one-time set-up costs, and with low implementation costs, this system would be renewable and cost-efficient energy because of no operating costs involved.