At the 2026 World Robot Conference (WRC 2026), Wang He, founder and CTO of Galaxy General, shared his insights with the media, emphasizing that while the humanoid robot industry is actively preparing for home integration, a critical obstacle remains. Although these robots have already been deployed in outdoor guidance roles and assisted-living facilities where they lead elderly residents in exercise routines, achieving the level of safety required for everyday household use is a non-negotiable prerequisite.
Wang He pointed out that ordinary household customers typically lack the technical expertise to manage or halt a robot in an emergency, making it imperative for the industry to prioritize physical safety measures. He acknowledged that there is still significant progress to be made in ensuring embodied physical safety before these products can be confidently introduced into homes. The ultimate goal, he noted, is to deliver robots that are both effective and entirely safe for everyday users.
During the conference, Galaxy General unveiled its latest innovation, the "Galaxy Xingzai," a compact humanoid robot equipped with agent-level learning capabilities. This robot stands out due to its lifelike movements, real-time motion imitation, and ability to interact with real-world environments. It can learn by observation and make autonomous decisions, positioning it as a strong candidate for future home applications.
Wang He explained that the robot's self-learning ability is supported by a "universal cerebellum," a brain region developed within the Galaxy General "Galaxy Brain" system. This cerebellum has been trained on over 100,000 hours of human full-body motion data, sourced from both high-precision motion capture devices and vast amounts of internet video content. While the internet-sourced data may not match the precision of motion capture, its sheer volume provides a comprehensive foundation.
Through simulation and reinforcement learning, this data has been synthesized into a versatile cerebellum capable of tracking and replicating human actions with remarkable accuracy. Wang He drew an analogy to remote-controlled robots used in hazardous environments like fire decontamination, explaining that the "Xingzai" applies a similar principle to dancing. By extracting the movements of a street dancer in real time through visual inputs, the robot can internalize and reproduce the entire routine seamlessly.
This development underscores Galaxy General's commitment to advancing humanoid robotics, with a clear focus on bridging the gap between industrial applications and the safe, practical use of robots in everyday life.