China’s recent pushes — from the World Humanoid Robot Games to high‑profile company listings — have placed humanoid robots in public venues such as schools, exhibitions and military training discussions. That visible rollout is part demonstration, part real‑world testing. Below I compare what the robots actually do, why Chinese actors are putting them into schools and similar institutions, and offer a short practical checklist for anyone planning a pilot deployment.
What these robots can demonstrably do
Multiple sources reporting from the World Humanoid Robot Games and company demonstrations show consistent, verifiable capabilities and limits:
- Mobility and agility demonstrations: robots ran, performed parkour‑style moves, backflips and kung fu routines in staged events. Reports note examples such as a robot called Tiangong Ultra completing a 100m sprint in 8.64 seconds and other machines scaling walls or executing acrobatics. These are public spectacle and engineering demonstrations rather than proof of general autonomy or broad usefulness (The Diplomat; TechSpot; New York Times).
- Task demonstrations aimed at service or industrial roles: at the Games some robots performed household chores and industrial tasks. However, multiple analysts and reporting emphasise these are narrow, scripted demonstrations rather than evidence of reliable, general‑purpose manipulation or long‑duration autonomy (The Diplomat; CNBC; Reuters as cited by TechSpot).
- Research and education use: Unitree and other Chinese firms derive a large share of early revenue from research and education customers; one filing cited research/education as nearly three‑quarters of humanoid revenue for Unitree in early periods. That supports the idea many robots are used as teaching and lab platforms rather than plug‑and‑play commercial workers (CNBC).
What the robots cannot yet do reliably
Across reporting there is agreement about clear technical constraints:
- Limited manipulation quality: robotic hands and manipulation remain imprecise and fragile for sustained, varied real‑world tasks; companies themselves warned in prospectuses that hands were not yet durable for broad adoption (CNBC).
- Short to moderate battery life and energy limits: analysts told reporters many humanoid models run for up to four hours while idle; energy consumption constrains long, independent missions and complicates field deployment (CNBC; TechSpot).
- Fragile control and fall risk: footage and eyewitness reporting from competitions showed frequent falls and control failures; observers pointed out these expose weaknesses in balance and control systems when robots encounter unexpected conditions (The Epoch Times; The Diplomat; TechSpot).
- Task‑specific training and brittle autonomy: experts stress robots must be trained for each task and still lack the intuitive, generalist ‘‘humanoid brain’’ to make rapid, flexible decisions outside scripted contexts (CNBC; The Epoch Times).
Why China is placing humanoids in schools, games and training grounds
Four practical motives appear across reporting — not mutually exclusive and supported by different sources:
- Public testing and iterative improvement: public events and school or research deployments provide large numbers of controlled, repeatable trials to refine hardware, software and human‑robot interaction in semi‑real settings (The Diplomat; CNBC).
- Education and workforce development: universities and labs use lower‑cost research models (e.g., Unitree’s EDU variants) to teach robotics, AI and manipulation — building a pipeline of engineers and realistic use cases (CNBC).
- Industrial strategy and commercialization pathway: China’s industry focus and subsidies aim to move robots from spectacle to factory, logistics or care roles. Demonstrations help sell the vision to investors and customers even while commercialization remains early (The Diplomat; CNBC; The Epoch Times).
- Military familiarisation and concept development: reporting of PLA interest and procurement notices shows defence actors are evaluating humanoids for reconnaissance, logistics and training scenarios; this is experimentation and planning rather than fielded combat systems (TechSpot; Reuters cited by TechSpot).
Evidence comparison: spectacle vs. testbed
Comparing accounts clarifies where demonstrations end and meaningful testing begins:
- Spectacle: World Robot Games footage and choreographed stunts (kung fu routines, backflips) show capability at a highlight‑reel level and serve public relations and investor signals (New York Times; The Diplomat).
- Testbed value: schools, research labs and educational sales are cited as the source of substantial early revenue and provide repeatable, documented interactions that inform sensor, control and manipulation engineering. They are where weaknesses—falls, manipulation limits, battery constraints—become quantifiable problems to solve (CNBC; The Epoch Times).
Practical checklist for deploying humanoid robots in a school or research setting
This checklist synthesises the documented capabilities and failure modes into concrete steps institutions should follow before accepting robots for demo, teaching or pilot work.
- Define goals precisely: research, teaching, public demonstration or pilot operations? Don’t conflate spectacle with validated use cases.
- Request hardware and software specs: battery runtime (reported models often run up to four hours idle), manipulation payload and durability, sensor suite, and supported autonomy level.
- Insist on staged acceptance testing: scripted demos, then randomized perturbation tests (uneven surfaces, unplanned obstacles) to evaluate fall risk and recovery.
- Plan safety and recovery: padded barriers, automated kill switches, and trained staff to intervene after falls or system errors; footage shows frequent falls even at high‑profile events.
- Budget for maintenance and training: expect higher R&D and spare‑part costs; early manufacturers derive much revenue from research/education models, which require hands‑on upkeep (CNBC).
- Data and security review: if connected devices are involved, conduct cybersecurity and data‑handling assessments; U.S. and European regulators have highlighted security concerns with some Chinese models (New York Times reporting on regulatory scrutiny).
- Set realistic timelines: treat deployments as multi‑year R&D pilots, not instant productivity gains. Experts describe humanoid progress as a marathon, not a sprint (The Epoch Times).
How China’s approach compares internationally
Reporting indicates China leads in unit volumes, manufacturing scale and public visibility, while Western companies emphasise different commercial paths:
- Scale and price pressure: analysts cited large shipment shares and falling average prices — one firm estimated a 93% price drop in average humanoid prices from 2020 to 2025 — which helps place research models into schools and labs cheaply (CNBC).
- Different public framing: Western coverage often treats Chinese displays as spectacle or state messaging; Chinese outlets emphasise industrial and social uses. Both perspectives are present in the evidence and shape deployment choices (The Diplomat).
- Regulatory and security friction: the U.S. has recently moved to restrict imports of certain Chinese humanoid and quadruped robots, a factor institutions should consider when buying or using these platforms for teaching or research (New York Times; CNBC).
Bottom line for U.S. schools and labs
China’s decision to put humanoid robots in schools and public events is a pragmatic mix of demonstration, testing and talent building. The machines shown can perform impressive, narrow feats of mobility and scripted chores but retain clear limits in manipulation precision, energy endurance and robust autonomy. Institutions considering pilots should treat them as research platforms: demand specifications, design for safety, budget for maintenance, and set multi‑year development expectations rather than immediate operational returns.
Citation anchors: “World Humanoid Robot Games” (5); “Tiangong Ultra” (5); “Unitree” (2); “research and education” (2); “hands were not yet precise or durable” (2); “run for up to four hours” (2); “falls and control failures” (6).