Autonomous-Car-Driving-using-CNN

Introduction

This project aims to develop a simulation environment for autonmous car driving using Convulutional Neural Network and Udacity Simulator. While Reinforcement Learning is the go-to functionality while dealing with decision making entities, we have chosen to go with Convulutional Neural Networks as they are light-in-weight when compared to Reinforcement Learning. Moreover, CNN provides better feature extraction

Requirements

Software Requirements

Python version 3.9.x

Udacity Simulator

Visual Studio Code

Library Requirements

Library	Jupyter / Cullab Installation	Command Prompt
os	!pip install os	py -m pip install os
numpy	!pip install numpy	py -m pip install numpy
matplotlib	!pip install matplotlib	py -m pip install matplotlib
tensorflow	!pip install tensorflow	py -m pip install tensorflow
keras	!pip install keras	py -m pip install keras
scikit-learn	!pip install scikit-learn	py -m pip install scikit-learn
imgaug	!pip install imgaug	py -m pip install imgaug
opencv	!pip install opencv-python	py -m pip install opencv-python
pandas	!pip install pandas	py -m pip install pandas
ntpath	!pip install path	py -m pip install path
random	!pip install random2	py -m pip install random2
eventlet	!pip install eventlet	py -m pip install eventlet
Flask	!pip install Flask	py -m pip install Flask
base64	!pip install pybase64	py -m pip install pybase64
io	!pip install python-io	py -m pip install python-io
PIL	!pip install Pillow	py -m pip install Pillow
socketio	!pip install python-socketio	py -m pip install python-socketio

How to Run

Installations

• Install the Udacity Simulator by the provided link. A zip file will be installed, which has to be extracted in the working directory

Training

• Open the Udacity Simulator and click on "play" on the displayed dialogue box. Click on "Training Mode". When the map is opened, click on the record button to record the images and their corresponding steering angles. This data acts as the training dataset, so it is advised to be careful while creating this dataset.

Virtual Environment Creation

• You can create the virtual environment in the anaconda prompt

Cloning

• You can clone the repository by executing this command in the gitbash environment in the required directory
git clone https://github.com/AAC-Open-Source-Poul/Autonomous-Car-Driving-using-CNN.git
• Make sure that all the paths are properly configured.