At the 2026 World Robot Conference (WRC 2026), Galbot unveiled the "Galaxy Xingzai," a humanoid robot touted as the world's first to possess agent-level learning capabilities. This compact humanoid features lifelike, agile movements, real-time motion imitation, and authentic scene-based interaction, marking a critical leap in embodied intelligence from "pre-set trajectories" to "learning while observing and making autonomous decisions."

According to Galbot founder and CTO Wang He, the core of Xingzai's breakthrough lies in the company's self-developed "Galaxy Brain" AstraBrain-WBC system. The latest 1.0 version has been fully deployed, enabling the robot to perform high-difficulty maneuvers such as handstand push-ups and synchronized street dance routines.

Behind these capabilities is over 100,000 hours of human motion data training, incorporating high-definition motion capture and internet video. This extensive dataset equips Xingzai with the ability to create real-time mappings from visual perception to motor control. During the live demonstration, Xingzai not only extracted and imitated a dancer's rhythmic body movements in real time but also interacted naturally with attendees in a real-world environment—initiatively handing out candy, recognizing terms like "beautiful sister" and "handsome guy" to offer personalized greetings, and even improvising interaction plans based on the surrounding context.

This "agent-level" autonomy breaks the traditional limitations of robots that depend on fixed commands or pre-programmed paths. Galbot stated that Xingzai has already evolved from learning physical movements to cognitive decision-making, with future plans for autonomous motion planning, scene imagination, and continuous learning abilities. "This is not just a slogan; it is a reality unfolding right now," Wang He emphasized. He added that all of Xingzai's demonstrations were genuinely recorded, and its showcased control precision, interaction fluency, and intelligent generalization capabilities open new possibilities for embodied intelligence to enter complex settings such as homes, service industries, and education.