AI-Driving-Car

Overview

This project is based on creating a autonomous car which is capable of sensing its environment and moving without human input. Basically it consists of 3 ultrasonic sensors attached to the front, left and right portions of car, these will calculate the obstacle distances infront of them and send them to the pretrained Machine Learning Algorithm. This will return a command, which is sent to the motors and move in the desired direction (Forward/Left/Right/Backward) accordingly.

Steps to reproduce

Clone this repo git clone https://github.com/maramsumanth/AI-Driving-Car.git, change to the current directory and run finalcode.py

Hardware Requirements

  Raspberry Pi          - 1
  L298N Motor Driver    - 1
  Bread Board (Small)   - 1
  LIPO Battery          - 1
  Chasis                - 1
  Power bank            - 1
  Wheels                - 2
  Motors                - 2
  Ultrasonic Sensors    - 3
  PS3 Controller for collecting data
  Jumper wires as needed

Data Collection and Model Training

Data is collected on different mazes using arduino, where the car is controlled using PS3 controller. It is stored in text file in the form Command, Sensor1, Sensor2, Sensor3, where each value is seperated by lines. Make sure to filter out the Stop command and it's corresponding sensor data in the dataset before training. Using python code this dataset is converted into sensor and command files. Finally it is trained using Random Forest Classifier(Machine Learning Algorithm), with accuracy of 93%.

Sample maze

Setup (For Linux)

Make sure that all these files are in the memory card of raspberry pi. If your Pi and laptop are on the same wifi network you can run code on Pi from your laptop using the command ssh username@ipaddress

Here is the drive link to video demonstrating the assembly of the components and another video containing the final output of project.