Meet GALBOT ET1

The Galbot ET1 is a bipedal humanoid robot developed by Beijing-based robotics company Galbot, unveiled in August 2026 at the World Robot Conference (WRC).

Key Features

Real-Time Learning: Powered by the Astrobrain-WBC (whole-body control) model, the ET1 can observe human movements and reproduce them instantly rather than relying on pre-recorded routines.
Training Data: The motion system was trained on approximately 100,000 hours of human movement data to produce smooth, life-like actions.
Bipedal Mobility: It is Galbot’s first fully walking, bipedal humanoid robot, shifting away from their earlier wheeled platforms like the Galbot G1.
Design & Safety: Standing roughly four feet tall, the robot features articulated feet for balance and safety features such as anti-pinch joint fabric. [6]

The Galbot ET1 (nicknamed “Galaxy Star”) is a groundbreaking, bipedal humanoid robot from Beijing Galbot AI Co., Ltd. that debuted at the 2026 World Robot Conference (WRC). It represents a massive leap forward in physical AI because it learns, tracks, and mimics human motion in real time without any pre-programming. [1, 2, 3, 4, 5]

Key Technical Specifications

Physical Form: It stands 173 cm tall, weighs 65 kg, and features an aluminum alloy skeleton enclosed by lightweight polymer panels.
Dexterity: Features a massive 48 degrees of freedom (DOF) across its body and 24 DOF in its hands, complete with five distinct fingers on each hand for high-precision physical manipulation.
Actuators: Driven by quasi-direct drive brushless servomotors paired with planetary and harmonic gearboxes.
Computing: Powered by the heavy-duty NVIDIA Jetson Thor compute system. [2]

The AI Brain: Astrobrain-WBC from Meet GALBOT

Instead of relying on rigid, pre-recorded engineering routines, the ET1 operates on Galbot’s proprietary Astrobrain-WBC general-purpose cerebellum model. [5, 6]

Autonomous Tracking: It follows the motto “I See. I Learn. I Act.” by instantly extracting human movement trajectories visually.
Massive Data Foundation: Its motion control system is trained on 100,000 hours of human movement data, delivering human-like, fluid coordination. [1, 5, 6]

Performance Capabilities & Real-World Goals

To showcase its dynamic balance and rapid learning, Galbot debuted the ET1 alongside a human street dancer. It successfully mirrored complex floor exercises, handstands, fluid transitions, and fast choreography on the fly. [5, 6]

While dancing is a flashy proof of concept, the real-world goal for the ET1 is deployment in manufacturing environments, commercial stores, and industrial workspaces. It is designed to navigate changing situations, interact with its environment, and master complex manual tasks just by watching human workers. [2, 5, 6, 7, 8]

If you want, I can dive deeper into: