Behavioural Cloning Project

Cloning simulated human driving behaviour using a CNN

INTRODUCTION

In this project, I used Keras and TensorFlow to implement a convolution neural network that was trained to clone human driving behaviour captured in a simulator. The network was then used to autonomously predict steering angles in the simulator to successfully drive the car around a track. A more detailed write-up is included in this project.

This project was undertaken as part of the Udacity Self-Driving Car NanoDegree.

HOW TO USE

Project dependencies

You can follow the guide explained here to help setup a working environment.

• https://github.com/udacity/CarND-Term1-Starter-Kit

Jupyter Notebooks

The included notebook visualises the dataset/model.

• analyse_model.ipynb

RELEVANT LINKS

Project writeup

• writeup.md

Original project repo

• https://github.com/udacity/CarND-Behavioral-Cloning-P3

RELEVANT FILES

• model.py contains the script to create and train the model
• model.h5 is my trained convolution neural network
• video.mp4 is a demo of the simulation using my trained model