🌎 embody
Demo | Video Tutorial | Paper
embody is a sensor and actuator environment for training digital consciousness (also known as artificial intelligence) algorithms.
This project is a work-in-progress and does not completely work as suggested by this document. Parts of the system work as suggested, other parts are still being worked on.
About
Embody is a set of interfaces to sensors and actuators on a computer. The interfaces provide a way to receive or update data about the state of a computer.
Sensor interfaces retrieve data about the state of a computer. For example: a screen capture video stream, mouse movement events and keyboard press events. Actuator interfaces update the state of a computer. For example: moving the mouse and pressing a keyboard button.
Benefits
💪 Flexible: Learn from across any number of sense, action and interface configurations
Sensor Features
👁️ Sight: See the world through a screen capture video stream, webcam or camera👂 Hearing: Use a microphone to listen to the world (WIP)🖐️ Touch: Feel through heat, texture and force feedback sensors (WIP)👃 Smell: Respond to air quality sensors to model scent (WIP)👅 Taste: Respond to water quality sensors to model taste (WIP)
Actuator Features
🖼️ Imagination: Create images on a 2D canvas by painting with pixels (WIP)🙌 Movement: Move in the world using a mouse, keyboard or motor👄 Speech: Create voice patterns to be played by speakers (WIP)
Limitations
🧪 Experimental: This project is a work-in-progress
Screenshots
Installation
Install the library
npm install embody -g
Example Usecases
API Reference
Documentation
To check out project-related documentation, please visit docs
Contributing
Feel free to join in. All are welcome. Please see contributing guide.
Acknowledgements
Learn More
Related Work
Some well-known artificial intelligence training environments include: Conceptual Learning (VicariousAI PixelWorld), Reinforcement Learning (OpenAI Gym, DeepMind Lab, garage), Algorithm-Agnostic (ROS, Embody)
License
MIT