X Square Robot Introduces Wall-B Embodied AI Model for Home Use

In a press release, X Square Robot announced Wall-B, an embodied AI foundation model for home robots, with initial deployments planned within 35 days. The company, backed by Alibaba, ByteDance, Xiaomi, and Meituan, introduced the model at an event in Beijing under the theme "Born to Bot, Bot to Family."

Wall-B is built on the company's World Unified Model architecture, which integrates vision, language, action, and physical prediction into a single training framework. X Square said this approach enables robots to operate in unpredictable household environments where tasks and layouts change frequently.

The model uses real household data to expose the system to varied scenarios such as misplaced items and unexpected obstacles. It also includes a physics-aware prediction mechanism that allows robots to anticipate physical outcomes before acting. During a live demonstration, a robot arranged flowers while adjusting its grip and movement in response to shifting stems without pre-programmed paths.

X Square acknowledged that the technology is still developing, noting that current systems sometimes require human intervention. The company stated that continuous operation in real-world homes will generate new data to accelerate learning and improvement ahead of broader deployment.