Adaptive Traffic Signal Time Controller

Abstract

The below solution is proposed to solve the problem of unnecessary waiting time at traffic signals. The timings of the traffic lights in the present scenario are fixed or hard–encoded and can be made dynamic through Artificial Intelligence and Deep Learning. The red light and green light times can be calculated based on the relative vehicle density at each lane. Our solution uses YOLOV7 in vehicle detection and an algorithm designed for the dynamic computation of timings.

Problem Statement

We all, at one time or another, have faced a condition through which the lane we are going in contains considerably higher volume traffic and the opposite or side lane contains very little. But due to fixed traffic signal timings, we have to wait a more extended time. This is inefficient management as the signal remains green for the other lane with no vehicles to pass. What if the traffic light timings, red and green were calculated dynamically based on the traffic conditions at that instance?

“Constant traffic signal timings irrespective of the density in lanes leads to unnecessary waiting. Design a system to automatically regulate the time of traffic light.”

Solution

The proposed system can be divided into three parts:

1. Capture Images: Cameras mounted on each lane will capture images at regular intervals to update the traffic conditions.
2. Vehicle Detection: The captured images will be processed by the YOLO model to detect the number of vehicles in each lane. The model will return the count of vehicles present in each lane, which will be the input for our time allocation algorithm.
3. Time Allocation: The count of vehicles received from the YOLO model will be used by the time allocation algorithm to calculate and accordingly allot the green and red light time for each lane.

Implementation Steps

1. Capture Photos: Regularly capture photos of all lanes when the red light is about to turn green to get the context of maximum traffic density at that particular time.

2. Vehicle Detection: Use a pre-trained and fine-tuned YOLOV7 model to optimally count vehicles from the images.

3. Categorize Lane Density: Define categories of lane density based on vehicle count. For instance, if the vehicle count lies in the [x,y] range, it is medium traffic; if the count > y, it is high traffic; and if the count < x, it is low traffic. This can vary based on different road and traffic square sizes.

4. Compare and Map Categories: Use the vehicle count from step 2 to determine the category mapping from step 3.

5. Time Allocation Algorithm: Use the algorithm and formulas to calculate the red and green light timings based on the traffic density category.

Technologies

YOLOV7:

YOLO (You Only Look Once) is a state-of-the-art, real-time object detection system. YOLOV7, the latest version, improves on the speed and accuracy of vehicle detection. It efficiently processes images to detect and count vehicles in each lane, providing essential data for the dynamic time allocation algorithm.

Read More -> https://arxiv.org/abs/2207.02696

Future Scopes

1. Dynamic Real-Time Adjustment: If all cars have crossed the signal and there is still a green light while other lanes have a red light, change the light accordingly in real-time.

2. Emergency Vehicle Detection: Detect emergency vehicles in a lane and turn the green light for that lane to allow quick passage.

By implementing this adaptive traffic signal time controller, we can significantly reduce unnecessary waiting times at traffic signals, leading to more efficient traffic management and smoother flow on the roads.

Code: “will be out soon”

This solution aims to revolutionize traffic management by making it more responsive to real-time conditions, thus enhancing the overall efficiency of road usage.