This is the story of six engineering students from a college in Bhopal who wanted to fix the serious traffic congestion problem in their city.
Meet the team: Sudhir Jain, Mayank Gattani, Palak Dubey, Batul Zamin, Sumit Badola and Anubhav Gupta.
They took on the civic issue as part of their project. Here is how they went about it.
Their idea was to split the overall time of traffic lights according to the traffic density of vehicles on different sides of the square. By this the traffic reduces on a particular direction. To do this we have used the cameras (which are already installed on the squares) then by image Processing we find out the areal densities of the vehicle standing on different sides of the roads by some algorithm developed by us. Areal density is the area occupied by the vehicle on the road. Then ratio of this areal density is take and the overall time is split according to this ratio .For eg. let us assume that we have only one way road with a square and vehicles can only go from north to south and from east to west . now assume that the ratio areal density comes to be 0.8 this means that 80% traffic density is on north direction as compared to the east direction so if we have total time like 100 sec we will split it in parts like 80 sec for north and 20 sec for south .
Project cost for traffic control is zero rupee as we are using the camera installed on square and there computers for whole task.
Technical details : we have used Beagle Bone Black and Arduino in this project prototype .
The system that exists
Many cameras are install on many square
Which is controlled by central control room ,
Time of traffic light is controlled manually,
Timer changes according to traffic at different time.
People who are in central control room they made the Challan manually.
Bugs in current system
No provision of instant change in traffic lights .
There are only 2-3 people appointed to control all
cameras of city.
They have to zoom and see individual vehicle if they break the law and note license plate number.
Every time someone has to sit and watch at cameras.
How we worked
We have added a clause of minimum 20% of overall time for each direction no matter what the ratio is.
We took the ratio of areal density.
The side whose density is more we give them more time to pass through green light.
There is no requirement of manual work in central office
The system works automatic and more efficiently.
To calculate areal density we use image processing.
We took an image and image processing is done through BeagleBone black it is type of mini computer.
We did programming in python in BreagleBone black
Image processing and traffic lights timer both are controlled by BreagleBoad black.
Along with this we also work on license plate detection on MATLAB
With 98% accuracy
Cost and problems with implementation
Cost for implementation is very low or approx. equal to zero
They have computers and cameras. We only need to implement software what we created.
Problem that we face is only accuracy, we never get 100% accurate result in this.
In finding areal density it would also count some unwanted area that is not a part of traffic
In license plate detection accuracy depends upon colour of car.
In three person on bike, we also get wrong result when there is child with them and or any person wear a helmet.
Photo credit: Christian Weidinger