CAPABILITIES
It does a little more than you expect
Six capabilities interlock: vision notices you first, the servos turn the body toward you, conversation happens on-device, and the expression display writes out what it feels.
Vision
It sees you before it speaks
The camera runs inference on the device, detecting faces and hand keypoints. It lifts its head when someone approaches, its gaze follows a hand moving left and right — and no frame ever leaves the robot.
- Face detection and light recognition to tell owner from visitor
- Hand keypoint tracking mappable to wave, stop and other gestures
- Live preview so you can tune gaze and framing
- Everything computed on-device; images are never uploaded
Conversation
Back and forth, entirely local
Wake it and talk. Speech recognition, reply generation and speech synthesis all run on the device, and the answer streams out — change your mind halfway and just interrupt.
- Wake-word triggered, no button needed
- Streaming synthesis with low time-to-first-word
- Multi-turn context with configurable memory length
- Bring your own conversation service — keys stay on the device
Expressions
A face that speaks for itself
Dozens of built-in expression states, and you can design your own. The face editor breaks eyes and mouth into placeable pixel blocks with live preview.
开心Happy
爱心Love
犯困Sleepy
惊讶Surprised
生气Angry
兴奋Excited
A selection of built-in expressions. More arrive with firmware and MOD updates.
Motion · Light · Sensing
Making presence land in actual behaviour
This is not a screen that talks — it is something that answers with its body.
Dual-axis motion
±90° pan and ±30° tilt let it nod, shake, look around or tilt its head in thought. Motion follows the conversation automatically, and a MOD can take it over completely.
Programmable light
An RGB indicator that breathes or pulses with state. At night it becomes a low-breathing silent notification, or just ambience.
Always-on sensing
The 6-axis IMU recognises being picked up, put down, pushed or tilted; capacitive touch tells a stroke from a press. Each gesture gets its own reaction.
Privacy: images, voice and conversation history stay on the device by default and are never uploaded automatically. Whether to connect an external conversation service — and which one — is your decision in settings.
Want a new expression or a new reaction?
The face editor and block editor both run in the browser — a few minutes is enough to teach it something new.