https://github.com/vpanjeta/self-driving-car

A Program for the implementation of Udacity's nano degree on self driving cars
https://github.com/vpanjeta/self-driving-car

cnn cnn-keras computer-vision deep-learning deep-neural-networks self-driving-car udacity

Last synced: about 3 hours ago
JSON representation

A Program for the implementation of Udacity's nano degree on self driving cars

• Host: GitHub
• URL: https://github.com/vpanjeta/self-driving-car
• Owner: VPanjeta
• Created: 2017-06-19T08:43:00.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2017-06-19T09:12:04.000Z (over 7 years ago)
• Last Synced: 2024-01-08T02:11:00.373Z (10 months ago)
• Topics: cnn, cnn-keras, computer-vision, deep-learning, deep-neural-networks, self-driving-car, udacity
• Language: Python
• Size: 4.62 MB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Self-Driving-Car

This is the code for Simulating a self driving car inspired by Siraj Raval's video on youtube.

See the ipython notebook provided for more details.

## Overview

I will be using Udacity's [Self driving car sim](https://github.com/udacity/self-driving-car-sim) for generating data for training and using as a car simulator while running the script. 


Yu can download the pre-built simulator or download the Unity project from the repository provided above.

## Installtion

Install the dependencies using the configuration file given as :

```python

# Use TensorFlow without GPU

conda env create -f environments.yml 

# Use TensorFlow with GPU

conda env create -f environment-gpu.yml

```

Then activate the environment as 

```

source activate car-behavioral-cloning

```

Or install the dependencies using pip.

## Usage and running

I have provided a pretrained model which you can use. 


Run the script as 

```python

python drive.py pretrained_model.h5

```

Then run the simulator in autonomous mode and see the script drive the car, you can see the calculated steering angles speed and other data displayed on the console.

## Training

For training on your own model you will need the data folder with images. This might take a while, for training on my GPU it took 7 CONTINUOUS HOURS!!!

```

python model.py

```

will generate models at every epoch, each better than the previous. 


run the model as 

```python

python drive.py model-xxx.h5

```