Run Gemini 3.1 Flash Live on an open source robot.

Thor from Google DeepMind connects the Gemini Live API to Reachy Mini, an open source robot from Hugging Face and Pollen Robotics. Real-time conversation, vision, multilingual switching, music generation with Lyria 3, Google Search grounding, and body movement synced to speech, all running through the Gemini Live API.

What's covered: Setting up the Reachy Mini SDK and environment, integrating the Gemini Live API using Antigravity, configuring voices and system prompt profiles, enabling real-time weather via Google Search grounding, generating music with Lyria 3, and wiring function calls to physical actuators for synchronized movement.

You'll need a Reachy Mini, a Gemini API key, and the Pollen Robotics SDK to follow along. Full project linked below.

Gemini API docs / Live API documentation / Pollen Robotics / Reachy Mini project / GitHub examples

What are you building with Gemini Live? Drop it in the comments.

Step-by-step tutorial →https://goo.gle/4cGkeKF 
Find the code →https://goo.gle/4soUjM9 

Chapters:
0:00 - Intro 
1:39 - Generating Moody Synth Music
3:18 - Multilingual Modes: German, French & Chinese
4:40 - Intro to the Open-Source Hardware & Gemini Live API
5:46 - SDK & Virtual Environment Setup
7:37 - Exploring the Codebase: System Prompts & Personalities
8:56 - Under the Hood: Lyria 3 Music & Search Grounding
10:08 - How Reachy Moves: Head Wobbler & Speech Sway
10:58 - Final Thoughts & Community Use Cases

Subscribe to Google for Developers → https://goo.gle/developers 

Speaker: Thor Schaeff 
Products Mentioned:  Google AI, Gemini

Google for Developers

A demonstration of integrating Google's Gemini Live API into a Reachy Mini open-source robot built by Hugging Face and Pollen Robotics. The robot, named Richie, showcases real-time voice conversation, multilingual support (90+ languages), vision via an onboard camera, weather lookup via Google Search grounding, music generation using Lyria, and body movement synchronized to speech. The setup involves installing the Reachy Mini SDK, activating virtual environments, and running a conversation app that passes a Gemini API key and model string. Custom tool integrations enable capabilities like room scanning, music generation, and animated head/body movements driven by audio buffers.

Give your robot a Voice with Gemini Live