Collision Guard

Introduction

This is an Android App based on flutter which will be used to detect collisions. This app leverages accelerometer sensors on the device which are used to send a text to emergency contacts in case it detects a collision.

Presentation Link

YouTube Presentation URL

Screens

Home Screen

As you can see, this is the home screen of the app.

• The first row of widgets namely X, Y, and Z are the accelerometer value in their respective axis.
• The widget in the middle is for the speed which updates as the mobile device moves.
• As soon as you click the button "Start Detection", the values are tracked.
• You can click "Show Driving Graph" to look at how your acceleration and speed were during your drive.

Graph Screen

This is the graph screen.

• This screen contains X, Y, and Z, accelerometer values alon with the Speed.
• This page can help us look back at our driving and maybe understand how this might have impacted the mileage.
• We also have an "Export data" button which will will export the sensors data to a text file.

Accident Screen

The following screenshots belong to the Accident? page.

• As soon as the page opens a 15 seconds timer starts.
• If you click I am safe within the 15 seconds, you go back to home page.
• If in case you do not and the timer runs out, emergency message is sent to your emergency contacts with your latitude and longitude information.

Settings Screen

These images above correspond to the settings screen.

• The first image is what you get when you select the settings tab.
• When you click Conacts, you get the second screen.
• In this screen you have the contacts you saved which include their names and contact number.
• You have an option to add a new contact by entereding their name and mobile number in the bottom.
• You also have an option to delete the contacts by sliding the contact to left.
• All these contacts are saved on Firebase Realtime database as shown below.

Implementation

In the flow chart above:

• When the app opens, the user has to click Start detecting. This is when the cycle begins.
• The accelerometer data is sent to the ML model deployed on Firebase continuously.
• As long as the prediction label is 0 (driving) we keep looping through the process.
• If the prediction is 1 (accident), the system looks at the GPS speed and if that is also 0 at that moment, the app starts the 15 seconds timer.
• If Stop timer is clicked before timer runs out, we go back to the home screen and detecting continues.
• If it is not clicked, the timer takes the contact data from firebase and sends all of those contacts a text message with your coordinates.

Steps to run (Method 1)

Prerequisites

• To run this file, you will need to install and setup flutter in your laptop.

• The emulator which was best tested for this app is a Pixel 7 Pro with API 33.

1. Clone the project

git clone https://github.com/kompellabhinav/collision-guard.git

2. Go to the project directory

cd collision-guard

3. Install the dependencies

flutter pub get

4. Run the flutter app

flutter run

Steps to run (Method 2)

Using apk

• Find the file named app-release.apk on the root foler
• Drag and drop the apk file on to the emulator.
• Open all apps on the emulator and open app names collision_detection.