---
title: "China Published a Humanoid-Robot Standard Framework — But It Is Not a Rulebook Yet"
date: 2026-11-17
category: Humanoids & Robotics
site: NeuroAI
canonical: https://neuroai.site/a/na-robot-standard-system-2026
language: en
---

# China Published a Humanoid-Robot Standard Framework — But It Is Not a Rulebook Yet

> A six-part framework maps what China wants to standardise in humanoid robots. It is a plan for standards, not an enforceable one.

Two factories build two humanoid robots. The arms use different connectors, the batteries speak different protocols, and neither maker's testing method matches the other's. A buyer who wants both has to pay twice for integration. This is the problem a new Chinese framework is meant to address — eventually.

On February 28, 2026, at the annual meeting of the Humanoid Robots and Embodied Intelligence Standardization (HEIS) technical committee in Beijing, China published its first national-level standard system for the sector: the Humanoid Robots and Embodied Intelligence Standard System (人形机器人与具身智能标准体系(2026版)). Xinhua and the State Council Information Office reported it as a top-level design covering the whole industrial chain and the full lifecycle.

## First, what this document is not

This distinction matters. The framework was **released**. It is not a mandatory national standard that has been passed and taken effect, and it is not a safety certification that any robot now has to hold. The committee said the next step is to organise government bodies, companies, institutes and universities to develop individual standards inside the framework.

In policy terms: the system has been published, the rules underneath it are still to be drafted. Anyone writing that China had "implemented" binding humanoid rules in February 2026 would be a stage ahead of the evidence.

## The six pillars

The framework is organised into six parts:

- **Foundational and common standards** — general definitions and guidance.

- **Neuromorphic and intelligent computing** — the "big brain, small brain" (大小脑) architecture, data lifecycle, and model training and deployment.

- **Limbs and components** — torsos, arms, legs, dexterous hands, actuators and perception modules.

- **Full-system integration** — complete machines, system software and hardware-software coordination.

- **Application** — how robots are developed, run and maintained in specific scenarios.

- **Safety and ethics** — requirements intended to run through the entire lifecycle.

That structure tells you where the industry's friction is expected: not in locomotion demos, but in interfaces, data and testing.

## Why a framework arrives now

The Ministry of Industry and Information Technology has described 2025 as China's first year of humanoid robot mass production, with more than 140 domestic manufacturers releasing over 330 models in twelve months. A market with that many models and no shared conventions is expensive to buy into.

Liang Liang, the committee's secretary-general, argued that a common system could unify technical specifications and evaluation criteria, reduce coordination costs across the supply chain and encourage modular components. Wang Xingxing of Unitree and Peng Zhihui of AgiBot, both deputy directors of the committee, used the meeting to show robots working in factories and to describe tactile sensing as a bottleneck that standardisation could ease.

The logic is familiar from other industries: agreeing on interfaces and test methods lowers the cost of buying from more than one supplier, which in turn makes scale possible.

## Who wrote it

The system was developed by more than 120 research institutions, enterprises and industry users, organised by the MIIT technical committee. The committee itself was established in December 2025. Its chair is Xie Shaofeng, MIIT's chief engineer; its secretariat sits with the Chinese Institute of Electronics. Company representatives make up more than half of its members, according to Chinese Youth Daily.

At the same meeting, the committee also published a preliminary list of priority standards projects and two industry initiatives — one on robot identity management, one on safe development. These are proposals and cooperation frameworks, not regulations.

That composition is worth noting. A standard-writing body dominated by manufacturers may move faster on practical specifications, but it also has to balance the interests of firms whose designs differ.

## Honest limitations

A published framework is a map, not a regulation. The sources I reviewed describe the six-part structure and the intention to develop standards; they do not provide the full text of the framework, a compliance timetable, or a list of which future standards will be mandatory rather than recommended. Any specific technical thresholds circulating in local government summaries should be treated as unverified until they appear in an official standard document.

The event and the quotes are reported by Chinese state media and state-affiliated outlets, which are reliable for what was said at the meeting but are not independent assessors of the policy's effect. I found no response from foreign manufacturers, no comparison with ISO or IEEE efforts, and no indication yet of how Chinese standards would interact with safety rules in export markets.

Finally, a standard system does not solve the harder commercial problem. It can reduce integration cost; it cannot create customers. Factory deployments in 2026 are still measured in pilot lines, and the framework does not change that.

## What readers can do now

- **If you buy robots:** ask vendors which of their interfaces and test results follow the new framework, and get the answer in writing. A framework is only useful if suppliers actually conform.

- **If you track policy:** distinguish the stages. Published framework, standards under development, approved standard, mandatory standard in force are four different things.

- **If you manufacture:** watch the committee's next output — the individual standards, not the umbrella document, are what will affect product design and certification.

---

Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-robot-standard-system-2026
Free to quote with attribution and a link to the original.
