HAL 9000

• 

  HAL 9000

• 

  Playing Chess with HAL

• 

  Internals

• 

  BeagleY-AI with Zen Hat

Inspiration

HAL 9000 (2001: A Space Odyssey)

What it does

Simulates HAL 9000 from the movie 2001: A Space Odyssey. Features include playing chess against HAL, using voice commands to speak to HAL/specify chess moves, and a chess GUI.

How we built it

Back-end in Python, front-end in Flask. Hosted on Beagle-YAI system. Details:

• Speech-to-Text: Google speech-to-text (facilitated by SpeechToText Python library).
• Custom parser for natural language to SAN (Standard Algebraic Notation) for chess.
• Chess engine: python-chess library.
• Chess AI: Stockfish.
• Chess GUI: Flask (using static chessboard SVG).
• HAL Voice Lines (Script): Gemini.
• Text-to-Speech: Elevenlabs API (custom-built HAL voice)

Challenges we ran into

• Setting up Beagle-YAI environment and audio devices.
• Inexperience with Flask front-end.
• Audio recognition issues (speech-to-text).
• Parsing natural language into chess notation.
• Coordinating multiple AIs and API calls (Google speech-to-text, Stockfish, Gemini, ElevenLabs).
• Multithreading coordination (UI and voice listening must run in background).

Accomplishments that we're proud of

• (Mostly) working voice commands.
• Coordination of a complex data pipeline despite being new to embedded systems.

What we learned

• Flask.
• Data pipelining.
• Parsing.
• Embedded systems.
• Audio.

What's next for HAL 9000

• Multiple states (eg Conversation Mode).
• Smoother startup/termination.
• More movie accurate (moral ambiguity of the character).

Built With

• beaglebone-yai
• elevenlabs
• flask
• gemini
• python
• python-chess
• speechtotext
• stockfish
• whisper