One problem that anyone driving in a city has encountered is traffic. Cities are renowned for their traffic and how long it takes to get from place to place due to the volume of people attempting to get places in such a small area. My discipline is computer engineering and I believe that some of the techniques that we apply could help address and decrease traffic in the city. Computer engineering can be thought of as a combination of computer science and electrical engineering. These skills can be applied to so many different things throughout a city due to computers and electronics controlling everything from energy consumption to communication. I think that an interesting application could be to control traffic signals to decrease commutes throughout the city. There is one computing system that I would utilize in traffic signal timing that could decrease waits and that would be to use machine learning, specifically an artificial neural network, to perfect waiting times at signals.
An artificial neural network is a form of machine learning modeled after the central nervous systems of animals. These networks take in sets of data and adjust their outputs based on current conditions and historical data. Each implementation of a neural network is unique in that you can control the weight of the inputs. It can give heavier weight to historical data or more recent data.
Currently, the traffic signals in Boston are set to timers. Depending on the time of day, the traffic signals will wait a certain number of seconds to change colors. According to the City of Boston website, the city retimes traffic lights roughly once every five years and has computer control of 544 out of 845 traffic signals. In order to implement a central control of the lights, a set of sensors at many different locations in the city would need to be installed. The inputs from these sensors would be fed into the neural network which could then control the timing of the traffic signals to be more efficient. The neural network also ‘learns’ from previous data. Based on its performance in previous situations, it can optimize the wait time of cars in the future. The traffic in different parts of the city could affect the timing of the lights half a mile away in order to increase efficiency. Quicker commutes lead to more time spent with family and enjoying their time rather than sitting in a car. Although this obviously wouldn’t completely solve Boston’s traffic problems, I believe that it would be fairly effective and increase the quality of life and decrease fuel consumption of those in and around the city.
