Shuttle tracking app to track shuttle inside a university/college campus without the need of GPS i.e., using static sensors along the route. Additional capabilities include delay and arrival time prediction.
Report Bug
·
Request Feature
The proposed system consists of three main modules. First module is the IOT data collection module which consists of a microcontroller, Arduino along with sensors to detect the presence of a shuttle. Sensors like ultrasonic and colour sensors can be used to detect when a shuttle passes, its location, timestamp, etc. The detection sensors will be placed along the route of the movement of the shuttle at a convenient interval like shuttle stops, for example every 200-300 meters. For fetching the data from the microcontroller, we have used the Bluetooth sensor HC-05. We could also use an internet module or a microcontroller embedded with internet module such as Node MCU, but for demonstration and prototype version, we have used just a Bluetooth sensor.
The second module consists of the User app which is made using React Native and supporting libraries. It fetches the data from the Bluetooth sensor on the Arduino using classic Bluetooth package which supports the Bluetooth version of the HC-05 sensor. The app displays information about where the shuttle is and which direction it is going in on the designated shuttle route map which is similar in appearance to metro route map. In addition, a notifications panel displaying a comment on the status of the delay time of the shuttle. The app sends the data to a backend server when the shuttle passes through 2 stops, sending the data for time between the 2 stops, day of the week and hour of the day for predicting the running status of the shuttle.
Third module is the server running on Node JS framework. A machine learning model has been developed using the Random Forest algorithm programmed in python language on a semi-generated dataset containing attributes like time between stop, day of the week, hour of the day, and the label being if the bus is late (1) or not (0). Datatype for each attribute is numerical. The model is trained on the dataset and saved as a .joblib file. Now, the app sends the data to the server using POST request and the server makes a prediction using the saved model. The prediction is either 0 (not late i.e., fast) or 1 (late i.e., slow). The data is sent back to the server as the response for the post request and the app displays the notification if the bus is faster or slower than usual.
This project is built with:
- React Native
- Ignite
- Python Server
Siddharth Pal - @originalsidd_ - originalsidd@gmail.com
Project Link: https://github.com/originalsidd/shuttle-app-frontend