Sustainability Consultant
AUTOMATED TRAFFIC LIGHT SYSTEM (ATLS)
Business and Implementation Plan
Tired of traffic? So was my friend and I. Being from Houston, I would find myself stuck in traffic in the heat of summer with my AC shutting off. After complaining and talking to myself, I started to notice slight improvements a city like Houston could make. Feel free to browse and let your curiosity wonder, what if.
OVERVIEW
Inspired by a resident in one of the largest growing cities in the United States, the Department of Transportation should consider a web-based automated traffic light system.
The purpose of this project was to not uproot the existing infrastructure by expanding roadways or tear up streets by placing an inductive loop under them. Our reason was to utilize the existing resources like the cameras on top of nearly all Houston's 184,000 traffic lighted intersection. These cameras would be linked to a deep learning software program that would be able to better predict the traffic patterns and behaviors whereas, the police officer standing next to the utility box could not only monitor the intersection remotely but all the intersections in its proximity.
Urban development starts with smarter infrastructure and planning through the adaptation of new technologies.
CHALLENGE
Developing a deep-learning algorithm is not easy. Especially for someone that shares only the fundamental skills in Python. It takes an extensive amount of time and research. Most of the research findings were found in China as they seem to be all-in on computer vision accompanied by a deep-learning algorithm.
The basis of artificial intelligence was built on the premise of machine learning which evolved into deep-learning. Your average PC might be able to conduct machine learning programs but deep-learning involves a systematic yet complex web of convolutional neural networks. These convolutional neural networks are programmed to identify the values you wish to measure and better yet, predict. Predict as in, each neural network serves as a filtering system to identify an object. This object feeds through the next neural net until it's able to process what that object is. Countless revisions in code and testing are involved to reach the optimal accuracy of ninety-eight percent efficiency.
One might believe there is no difference between a percent. But the fractional jump between percentages is a milestone and after each simulation, these percentages tend to move upward or downward based on the quality of the images or malfunctions with the code or convolutional neural network. This program is similar to a newborn, where it is processing information for the very first time attempting to distinguish and associate it from all of its surroundings.
Attempting to predict behaviors and trends can take years without reaching optimal efficiency. Let alone the level of mathematics and physics involved to develop the algorithm. As you can assume, it is quite a difficult process but you might ask, aren't their mathematicians and physicists that can figure this out? Sure. But the term artificial intelligence has only been understood by experts for no more than thirty years. These experts have concentrated knowledge in the subject matter where mathematicians and physicists lack.
The time, energy, repetition, and knowledge to make an artificially intelligent program come to life is no small feat.
For the preliminary stages of artificial intelligence to grow technological advancements for graphic cards and processors must evolve for these systems to become feasible in order to lower the operating expenses a team uses for data centers to test and run their programs. Otherwise, teams might spend a majority of their capital to configure their own data center.
VALUE
There is tremendous value in utilizing computer vision and deep learning. It is the future for many traditional forms of computer-based operations.
Municipal-run traffic operations will see the highest efficiency in terms of decreasing congestion and anthropogenic emissions. Public transit will see maximized performance rates, and commuters will share a higher quality of life with the psychological effects involved with traveling. Expanding roadways or developing new tunnels or overpasses is a costly expense that has proven to be effective only for a short duration after development is complete.
Imagine a time where traffic lights no longer operate on time intervals but on behavioral trends and patterns. A sizable investment for local municipalities and the federal government to lower their futures bottom-line expenses and increase their constituents top line. This future is taking shape on our highway systems and through our automaker's technological-based offerings. When we discuss autonomous vehicles, automated traffic light systems must be brought into the same discussion.
With the right infrastructure in place, smarter city grids have the potential to streamline the production feasibility of autonomous vehicles for the end-customer. Commercial trucking and public transit have the opportunity to see productivity and efficiency rates at an all-time high. Local law enforcement can be dispersed for enforcing the law rather than barricading intersections and timing lights. The value proposition is weighed on the technological advancements, feasibility, and return on the initial investment.