Behavioral Cloning

Overview

Develop a deep learning neural network to replicate (clone) driving behavior. To accomplish this a convolutional neural network (CNN) is developed to correctly steer a car along the road. Specifically the CNN will be trained using data from a specific track. Many aspects of Keras, OpenCV, python, numpy, and matplotlib are used to develop the CNN. The model can be found here model.py

Installing and Running the Classifier

The following files are included :

• model.py Keras implementation of behavioral cloning model
• sdc_lib.py Helper functions for the behavioral cloning model
• model.h5 Autonomous driving capture of the behavioral model
• run1.mp4 Video capture of autonomous driving of the behavioral model
• writeup_report Detailed writeup of model creation and evaluation
• sample_data_set A sample of images from the data set
• model_mse_loss.csv Keras CSVLogger training output
• simple_clone_model.py Very simple keras models for evaluating data set

The following steps are used to run the model:

1. Install miniconda environment and related packages

   https://conda.io/miniconda.html
   

2. Clone the SDC-BehavioralCloning git repository

   $  git clone https://github.com/jfoshea/Driver-Behavioral-Cloning.git
   

3. enable cardnd-term1 virtualenv

   $ source activate carnd-term1
   

4. Train the model (Note data set is needed to train a model)

   $ python model.py
   

5. Run the model in autonomous mode in the simulator.

   $ python drive.py model.h5
   $ Launch simulator select track and click Autonomous Mode
   

Writeup

A detailed writeup of the behavioral cloning project and challenges are located here [writeup_report] (https://github.com/jfoshea/Driver-Behavioral-Cloning/blob/master/writeup_report.md)