In 2020, the number of cars in the world reached 1.4 billion cars. Knowing this fact helps us understand why the streets are stuffed with cars in the vital places.
It becomes clear that the number of cars increases every year, making it harder to drive. How Many times do you arrive at your work late because of a jam? And how many times did you find yourself in a jam and can’t move your car at least one inch for hours?
Well, as long as this problem increases every year, we need a smart solution for it. A solution that can decrease the jam at a good rate.
Smart traffic lights
If we could improve the traffic lights’ system, we could solve the problem. And with the help of Machine Learning ML, we can teach the system to learn from data in order to take appropriate decisions.
A city in India called Visakhapatnam is taken for implementing an intelligent traffic signal controlling (ITCS). And for simplicity we can call it Smart Traffic Lights.
Applications of ITSC
This tool is created to observe and solve road problems. A simulation of SUMO was explained by a group of researchers in a paper of “Microscopic Traffic Simulation using SUMO”.
SUMO Is designed to handle large road networks. It’s also available as an open source package to easily use it.
Then, a group of two people from France used SUMO to create a realistic simulation model for road transport systems. Then, the information provided from this experiment was used by the Cooperative Intelligent Transportation System (C- ITS) to improve the road traffic.
Some researchers tried to reinforcement learning to road traffic control as it shows a great ability to solve complex road problems. The reinforcement learning uses in its algorithm the Deep Neural Network (DNN) and the Convolutional Neural Network (CNN) to apply in traffic control systems.
Step 1: Training the model
This step is done through training on the data we’ve collected before. Knowing that we can use the Support Vector Machine (SVM) Model in our program.
The SVM Model is one of the machine learning algorithms that can help classifying traffic signal junctions.
Step 2: Begin simulation
By running SUMO through TraCI which is a Python library used to access the running road traffic simulation in SUMO.
Step 3: Get traffic information
The TraCI library helps in the process of information retrieval at every traffic signal junctions.
Step 4: Classifying Traffic Signals
The classifying contains three categories. The categories for simplicity can be divided into low traffic, medium traffic, and high traffic.
The classification or the categorization part can be done through the ML model with a real and sufficient training data from SUMO. Then, it can label new observations and categorize them itself.
Step 5: Changing traffic signals of high-traffic areas
In this step we try to bring the high traffic areas to low or medium traffic. This can help a lot in decreasing jams in those areas.
Step 6: repeating the steps from 3 to 6 as much as we need in order to get good results.
Real Life Example:
A real life application for this technology is Google Maps. This app can define a route for you and detect jams suggesting other routes to take.