You can use this file as a template for your writeup if you want to submit it as a markdown file, but feel free to use some other method and submit a pdf if you prefer.
Behavioral Cloning Project
The goals / steps of this project are the following:
- Use the simulator to collect data of good driving behavior
- Build, a convolution neural network in Keras that predicts steering angles from images
- Train and validate the model with a training and validation set
- Test that the model successfully drives around track one without leaving the road
- Summarize the results with a written report
Here I will consider the rubric points individually and describe how I addressed each point in my implementation.
My project includes the following files:
- model.py containing the script to create and train the model
- drive.py for driving the car in autonomous mode
- model.h5 containing a trained convolution neural network
- writeup_report.md or writeup_report.pdf summarizing the results
Using the Udacity provided simulator and my drive.py file, the car can be driven autonomously around the track by executing
python drive.py model.pth
The model.py file contains the code for training and saving the convolution neural network. The file shows the pipeline I used for training and validating the model, and it contains comments to explain how the code works.
My model consists of a convolution neural network with 5x5 and 3x3 filter sizes and depths between 24 and 64 (model.py lines 8-11)
The model includes RELU layers to introduce nonlinearity (code line 20), and the data is normalized in the model using a pytorch torchvision.
The model contains dropout layers in order to reduce overfitting (model.py lines 16).
The model was trained and validated on different data sets to ensure that the model was not overfitting (code block 126). The model was tested by running it through the simulator and ensuring that the vehicle could stay on the track.
The model used an adam optimizer, so the learning rate was not tuned manually (model.ipynb block 126).
Training data was chosen to keep the vehicle driving on the road. I only used the colored center lane driving view.
For details about how I created the training data, see the next section.
The overall strategy for deriving a model architecture was to use the NVIDIA paper as a baseline and tweak from there.
My first step was to use a convolution neural network model similar to the NVIDIA self driving paper I thought this model might be appropriate because they both solve similar problems.
In order to gauge how well the model was working, I split my image and steering angle data into a training and validation set. I found that my first model had a low mean squared error on the training set but a high mean squared error on the validation set. This implied that the model was overfitting.
To combat the overfitting, I saved the model only when the validation errors increased
Then I used early stopping technique.
The final step was to run the simulator to see how well the car was driving around track one. There were a few spots where the vehicle fell off the track... to improve the driving behavior in these cases, I trained longer and grayscaled the images.
At the end of the process, the vehicle is able to drive autonomously around the track without leaving the road.
The final model architecture (model.py lines 20-31) consisted of a convolution neural network with the following layers and layer sizes.
To capture good driving behavior, I first recorded two laps on track one using center lane driving. Here is an example image of center lane driving:
I used the pytoch torch vision to perform some random horizontal flip with 0.7 probability and standardaization on mean 0.485 and std of 0.229. I also cropped out the image at the center. This random augumentation helped in preventing overfitting.
I finally randomly shuffled the data set and put 20% of the data into a validation set.
I used this training data for training the model. The validation set helped determine if the model was over or under fitting. The ideal number of epochs was Z as evidenced by ... I used an adam optimizer so that manually training the learning rate wasn't necessary.