At the opening day of CeMAT Asia 2026, the 26th international logistics technology and transport systems exhibition, ThunderSoft's Xiaowu Intelligence and the Beijing Humanoid Robotics Innovation Center jointly presented a full-stack industrial embodied AI platform. The platform, which combines multiple robot types under a single control system, aims to bridge the gap between warehousing and production lines in industrial logistics.
According to the Chinese outlet First Electric (D1EV), the platform integrates two major systems. The Beijing Humanoid's "Huisi Kaiwu" general-purpose embodied AI platform provides core capabilities such as embodied large models, cross-embodiment VLA (vision-language-action) foundation models, and whole-body motion control, handling long-horizon task planning and spatial understanding. Xiaowu's industrial embodied platform, built on an industrial-grade real-time kernel and business management system, enables coordinated operation and unified scheduling of multiple vehicle types, ensuring real-time closed-loop control for individual machines.
Three robot architectures, one scheduling system
At the exhibition, robots with three different motion architectures worked together under the same scheduling framework. The Beijing Humanoid's "Tianyi" series embodied robots performed precision tasks such as valve core assembly, screw picking, and bin loading/unloading. Forklift FMRs handled pallet-level heavy lifting and high-level storage, while latent-type AMRs (autonomous mobile robots) navigated between storage areas and workstations for shelf retrieval and bin transport. All tasks were orchestrated by the platform, achieving precise coordination between mobile transport and embodied manipulation.
The platform targets high-frequency pain points in industrial logistics and smart manufacturing, covering automotive manufacturing, intelligent logistics, semiconductor manufacturing, and 3C home appliances. It addresses end-of-line material delivery from stereoscopic warehouses to line-side storage, shelf-to-workstation distribution, and labor-intensive tasks like unordered picking and automatic loading/unloading, linking material flow into an unmanned continuous operation chain.
Since signing a strategic cooperation agreement in January 2026, the two parties have demonstrated heterogeneous robot cluster collaboration at events including the World AI Conference, the World Humanoid Robot Games, and the World Robot Conference. Going forward, they plan to continue delivering lightweight, mass-producible intelligent robot cluster solutions for automotive manufacturing, high-end equipment, and smart factories, promoting the scalable deployment of industrial embodied AI.
