This is a deep neural network to help the car self steer or as udacity calls it car behavioral cloning , the network is fed three images from 3 different angles where each 3 images are labeled with a steering angle
The model is built based on the Nvidia model, more details of this model can be found in their published paper.
I used the udacity simulator, for generating data and testing in the autonomous mode.
Note: This project was inspired by naokishibuya's project which is based on Nvidia's paper too.
See the final results See video
-
Currently both training and validation loss are about 0.04-0.05, which results in a pretty good result
-
Generating more data for the udacity simulator would definitely help
-
Data used here
Open the udacity simulator, choose autonomous mode and type the following command in your terminal:
python drive.py
Tensorflow 2 (with Keras interface).