Self Driving Car in TensorFlow!

Video tutorial link:

https://img.youtube.com/vi/6YC5p-r-Xhg/0.jpg

1. Clone this repo

2. Install Conda if not already installed (Python 3.6 version)

   https://www.anaconda.com/download/

3. Download and extract Udacity's Self-Driving Car Simulator

   Linux
   Mac
   Windows

4. Run Udacity's driving simulator in training mode

   When prompted where to store recorded data, create a folder called "data" in this cloned repo folder and save there.

5. Create conda environment

   Non-GPU users: conda env create -f cpu.yml
   GPU users: conda env create -f gpu.yml

6. Run model_tf.py to train. models are output to this repo folder

   python model_tf.py -N [A model name of your choosing]
   Run python model_tf.py --help for more options

7. Start Udacity's driving simulator and enter autonomous mode

8. Run python drive_tf.py [model_name] to connect model to Udacity's driving simulator

   Note: model name is xy...z.meta without ".meta"