Huawei Cloud and Fulongma Launch Sanitation Robot With Cloud Brain and Edge Cerebellum
On September 18, during Huawei Connect 2026, Huawei Cloud and Fulongma Group jointly held a product launch, unveiling a new-generation sanitation robot built on Huawei Cloud's CloudRobo embodied intelligence platform.
It is the first embodied intelligence product the two sides have released together since signing their cooperation framework, marking a shift from strategic consensus to product deployment for a model described as "Huawei Cloud builds the foundation, Fulongma defines the product."
End-to-end autonomous driving and embodied intelligence stacks span large models, compute and data closed loops — heavy-asset infrastructure that no single company can shoulder alone.
Huawei Cloud supplies the CloudRobo autonomous driving base, the CloudRobo embodied intelligence platform, Ascend device-cloud compute and the Octopus data closed-loop toolchain. Fulongma contributes a real-world data field covering more than 130 sanitation projects and over 15,000 connected devices, 625 patents, and engineering validation of L4-level products deployed in more than 40 cities.
The design pairs a cloud brain with an edge cerebellum. The cloud uses massive compute for global scheduling and complex decisions, while lightweight edge models deliver millisecond-level real-time response; the two cooperate bidirectionally to drive continuous model evolution.
According to the technical roadmap, the partners aim to complete end-to-end perception reconstruction in 2026, achieve GOD perception plus PDP planning-and-decision evolution in 2027, and fully integrate the CloudRobo embodied intelligence large model in 2028 — a generational leap from autonomous driving to embodied intelligence.
The core pain points in sanitation are inadequate cleaning and ineffective operation. The two companies will build a sanitation-specific world model to give robots three capabilities: deep task understanding, dynamic working-condition reasoning and intelligent anomaly handling — autonomously breaking down tasks, inferring the best strategy from real-time sensing and waste properties, and deciding safe responses to anomalies. For scattered construction waste, a robot can identify its high hardness and low adhesion and switch to high-pressure washing plus powerful vacuum sweeping; for rain-season clinging leaves, it can adopt a combined pre-wetting, softening, washing and sweeping strategy.
Huang Yingjun, Fulongma Group vice president and AI CTO, said the partnership with Huawei Cloud let the company skip three to five years of basic capability building and substantially shorten its R&D cycle by standing on top-tier industry compute and AI toolchains.
The logic is division of labor: Huawei Cloud handles compute, platform and toolchain, while Fulongma handles sanitation scenario understanding, equipment manufacturing and engineering deployment. Sanitation is a large-scale, repetitive, labor-intensive urban service, and while cleaning demands less of robots than humanoids do, the openness of the scene and the diversity of waste impose real requirements on perception and decision-making. Fulongma's more than 15,000 connected devices and 130-plus sanitation projects give the system real data sources and validation scenarios.