Intro

I took a course on soft computing (CE30501) in 6th sem. Our course instructor, Prof. Sudhir Kumar Barai (Former Dean UG (IIT KGP), Director BITS-Pilani) proposed us an interesting way to implement what we learnt in this course. He gave us 'smart campus' theme and ask us to come up with solutions relevant to the theme and present a working model in front of the class.

The title of our project was Bus route optimization. Upon presenting the model, our work was widely appreciated and can be implemented in reality.

Bus route optimization

January - March 2019

About My Project

How I did It

This project was given to our team of 5 batch-mates. Under the theme 'Smart campus', our team members pitched different ideas like smart attendance system, mess food wastage optimization and an idea suggested by me - Campus transport route optimization. After carefully considering available resources and time of completion, we finalized our project as 'Bus route optimization'.

The Problem statement: The buses of the IIT Kharagpur follow a definite path despite of the change in the weather condition or the demand. Optimize the bus route in real-time with maximum coverage of the bus stops and minimum expenditure of fuel.

The project can be broken down into the following part:

Data collection and insights
Fuzzy logic implementation
Djikstra algorithm and OpenCV implementation

My contribution to this project was data collection, fuzzy logic and OpenCV to predict and estimate the number of passengers.

Skills and Tools

Data analysis
Fuzzy logic and membership function
OpenCV
- Background subtraction
- Video alteration
MATLAB Simulink (for GUI)
Djikstra algorithm
MATLAB fuzzy toolbox

Data collection and insights

In order to proceed with the model, we circulated a google form with carefully selecting the parameters which affect the demand for buses.
After taking suggestions from students around campus we reached the conclusion that the following factors impact the most:
- Rain
- Temperature
- Special events like exam
Total of 517 entries was recorded. The insights from the data collected are shown below.
Some of the surprising insight we got was more than 90% of students preferred mode of transport os cycle

Fuzzy logic for demand estimation

After analysis, we found the following
- Cycle used - 91.3%
- Not in good condition - 8.7 %
  - Bus -4.4%; Walking - 3%; Others 1.3%
- Would take/prefer bus as a mode of transportation in the following cases
  - Normal day - 7.5%
  - Hot day - 54.10%
  - Rainy day - 69.40%
  - Exam - 21.90%
From the ERP we got the actual number of students leaving for classes according to the time table.
We then used the figure obtained above to estimate the demand at a particular time of the day. For ex. on Friday at 8 A.M. from LBS hostel 325 students leave for the lecture hall. Then on a normal day, 24 students (7.5%) will board the bus.
Next, we used fuzzy logic from MATLAB toolbox for three parameters: Rainy day, Hot day and Special event to estimate the demand at any time.

Djikstra algorithm and passenger counter (OpenCV)

The estimation of the number of students is made and then the data is fed into route planner.
The route planner is based on Djikstra algorithm.
GUI (Fig1) was made using MATLAB Simulink. It requires a number of bused available, the number of students at every node. Then it shows the optimized route for any given circumstances
We have also integrated a passenger counter using computer vision (Background subtraction) (Fig 5)
It recognizes the moving object and tracks it. If the object crosses the virtual lines drawn then it is counted as +1. (Fig 4)
Number of passengers hence obtained is used to further update the database and get accurate results.

Result

By following the proposed route annually, fuel equivalent to the amount 1 Lakh rupees can be saved annually. This project can be further integrated with IoT devices and app for better update and feedbacks. The project made into the top 5 projects presented.