XSquare's Single Model Masters Housework & Logistics, Beats Figure by 45%
机器之心 · wechat · 2026-08-20
Jiqizhixin reports on XSquareRobot's showcase at WRC, where its unified embodied model WALL-E (WALL-B) drives robots in both domestic chores (scooping cat litter, folding clothes) and industrial logistics.
Key Highlights:
- Logistics Performance: Achieved a sorting speed of 1,816 items/hour with 98% accuracy. This is 45% faster than Figure AI's publicly stated 1,248 items/hour, at only 30% of the deployment cost using dual arms + grippers.
- Home Service: Launched home services with 58 Daojia and the "X Family Member Plan" for long-term in-home deployment. Selected as an APEC case study.
Technical Architecture (WALL-B):
- Based on "World Unified Model" (WUM) architecture, integrating vision, language, touch, action, and physical prediction in one end-to-end network.
- Features native multimodal fusion, proprioception (sensing body position without full view), and understanding of physical laws.
- Supports interactive self-evolution, updating the model directly from failures without factory retraining.
Infrastructure & Moat:
- QUANXTAZero data collection scheme allows a 10:1 mix of non-embodied to real robot data to match full real-data performance, reducing data costs by 95%.
- Full industrial data pipeline from collection to deployment, shortening the dev cycle from 6 months to 3 days.
Business: Valuation exceeded 20 billion RMB, backed by 30+ investors including Xiaomi, Meituan, and Alibaba.
More from Embodied
- CMU's Andrea Bajcsy on what safety means for generalist robots, beyond collision avoidance — CSProfKGD · 2026-10-16
- Codex drives a physical CAN-bus motor, eyeing a fully Codex-built robot — OpenAIDevs · 2026-10-16
- Berkeley paper uses the Midas myth to rethink robotics beyond human imitation — begusgasper · 2026-10-16
- Team uses Claude Opus 5.5 to reproduce core claims of 10 recent robotics papers end to end — teortaxesTex · 2026-10-16
- Kinex: a harness that connects coding agents to any physical robot for scalable learning — daibond_alpha · 2026-10-16
- Gemini is a robotics outlier: dropping it lifts capability-robotics correlation from 0.20 to 0.96 — ycombinator · 2026-10-16