Persisted in data/motions/config.json Β· applies live via DDS
Manual sink / source override
Quick Actions
BUSY
Live Voice Commands
Last heard: --
Pending action: --
Audio attached: -- | Arm attached: -- | Gemini: --
Live Gemini Process
State: --
User: --
Gemini API Key
The key used by GeminiVoiceClient and the Live Gemini subprocess.
Saved to data/motions/config.json. Get a free key at
aistudio.google.com/app/apikey.
Typed Replay Engine
Wake Phrase Manager
Motion Control
BUSY
Replay Manager
Macro Recorder (Audio + Motion)
Voice (WAV)
Motion (JSONL)
Speed
Camera Vision & Face Recognition
--
--
Toggles take effect within ~1 second on the running Gemini session β no restart required.
Live Preview
Camera off β toggle Vision ON to see the live feed.
--
Resolution / FPS
JPEG Quality
Each button rebuilds the capture pipeline (~0.5 s). Modes match the
RealSense D435I colour sensor β on USB 2.x, stick to 424Γ240 or 640Γ480.
If the feed is grayscale/IR, pin the colour node with SANAD_CAMERA_USB_INDEX.
Add New Face
Tip: add 2β3 photos / different angles per person for best recognition.
The description is sent to Gemini with the photos β it can then greet
and talk about the person using what you wrote.