Skip to content

bakoushin/self-driving-car-javascript

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Run a self-driving car using JavaScript and TensorFlow.js

This project demonstrates how to train a self-driving car to steer and to drive autonomously in Udacity Self-Driving Car Simulator using TensorFlow.js.

Featured on TensorFlow YouTube channel:

See also detalied explanation of this project in a Medium publication: Run a self-driving car using JavaScript and TensorFlow.js.

Usage

  1. Clone this project.
  2. Download Udacity Simulator for Term 1.
  3. Record data using the simulator.
  4. Train the model using recorded data.
  5. Run the simulator in autonomous mode.
  6. Drive the car using the trained model.

Cloning and initialization

git clone https://github.com/bakoushin/self-driving-car-javascript.git
cd self-driving-car-javascript
npm install

Note: the code in this project is expected to run on Node.js 12 or greater.

Training a model

node train.js [--data DATA_DIRECTORY] [--model [MODEL_DIRECTORY] [--epochs NUMBER_OF_EPOCHS]

# Example:

node train.js --data ~/Documents/track1 --epochs 3

Default values:

  • data = 'data' directory within the project directory
  • model = 'model' directory within the project directory
  • epochs = 10

Driving a car

node drive.js [--model MODEL_DIRECTORY] [--speed SPEED_LIMIT]

# Example:

node drive.js --speed 20

Default values:

  • model = 'model' directory within the project directory
  • speed = 30

Author

Alex Bakoushin

License

MIT