From the Canton Fair to the Living Room: Haier W3 Consumer-Grade Exoskeleton Brings "Walking Assist" from Rehabilitation Centers into Ordinary Homes

[00:00] Recently at the Canton Fair in China, there were all kinds of products on display [00:03] One particularly eye-catching item was [00:05] A robot that helps you walk [00:07] An exoskeleton robot [00:09] So, like, I have elderly family members at home [00:12] I'm quite interested in this [00:13] Take a look at this video first [00:14] You'd never guess — wearing it, you can walk seven kilometers in one go without getting tired [00:17] The Haier W3 exoskeleton wonder [00:19] Completely blew my mind [00:21] Carbon fiber frame [00:22] Whole unit weighs only 1.8 kg [00:24] Not much heavier than a phone [00:26] "Imperceptible wear" is no exaggeration [00:27] Elderly users walk as if flying [00:29] Young users get an instant power-up [00:30] Even those with limited mobility can walk further with ease [00:32] Live try-on at the venue

Canton Fair Live: A Robot That "Helps You Walk"

If the public's impression of consumer-grade robots over the past few years has largelyremained in robot vacuums, quadruped robot dogs, and backflipping bipedal humanoids, then what truly went viral at this year's Canton Fair is a long-underestimated category — the walking-assist exoskeleton.

In the video, the host described the on-site experience with "you'd absolutely never guess that you could walk seven kilometers in one go without getting tired." The rhetoric naturally carries an advertising flavor, but the product itself is genuinely worth attention: it is no longer a behemoth locked inside rehabilitation centers, but a wearable device with a total weight of just 1.8 kg — billed as "barely heavier than a phone."

Haier W3 walking-assist exoskeleton, total weight approximately 1.8 kg, carbon fiber body, claimed to deliver roughly 7 km of continuous walking on a single charge. (Mentioned in the video)

More crucially, the host emphasized that "imperceptible wear is no exaggeration" — this is precisely the core threshold that separates consumer-grade exoskeletons from medical-grade products. If a piece of equipment, once worn, distorts the gait or delivers stiff joint torque interventions, then it is merely a "motorized brace"; only when the wearer almost forgets its existence does it deserve to be called consumer-grade.

Haier W3 Spec Teardown: How 1.8 kg Was Achieved

Pushing the total weight down to 1.8 kg is a fairly aggressive target for a lower-limb exoskeleton. From the video, we can pick out a few key design elements:

• Carbon fiber frame: The mangled phrase in the transcript is an ASR misrecognition of "carbon fiber skeleton." Carbon fiber is practically standard in hip-assist modules — it both guarantees structural stiffness and keeps the weight within a range users can actually accept.
• Hip assist as the core: Judging from the on-site footage and description, the W3 is a walking-assist exoskeleton built around hip assist, not a fully powered lower-limb unit. That also explains why 1.8 kg is feasible — it doesn't need a separate servo motor at the knee and ankle.
• Brushless motor + sensor fusion: The creator mentions "there's AI inside that adjusts based on your walking speed and posture," which means the machine relies on IMU and pressure sensors to capture gait phase in real time, then the algorithm outputs joint torque.

Key specs (mentioned in the video): 1.8 kg total weight, carbon fiber frame, hip assist, AI gait adaptation, ~7 km endurance per charge.

Worth noting, the creator leans on classic sales-pitch rhetoric like "seniors barely notice it, young people get a straight-up buff." But setting the wording aside, one detail worth recording from the — sorry, from the on-site demo — is that "the guy trying it on climbed three flights of stairs without even breathing hard." Stair climbing is one of the daily actions that demands the most from hip strength; if that scenario holds up, it means the torque output has genuinely reached a perceptible level of assist.

Price War: From "Hospitals Can't Afford It" to "Insurance Can Reimburse It"

The host's historical comparison of prices in the video actually pinpoints the single most critical turning point in the consumer-grade exoskeleton track:

"The difference before was that it was very expensive—it was meant for those kinds of rehabilitation centers... A hospital rehab center would buy one, help you through a period of time, and that was it. Ordinary people simply couldn't afford it. But now the price has been driven down, roughly in the seven-to-eight-thousand to twelve-thousand range, 11,999 RMB."

Translated into industry language: over the past decade, rehabilitation exoskeletons (medical-grade) carried unit prices that routinely ran into the hundreds of thousands or even over a million RMB, with buyers essentially limited to the rehab departments of top-tier hospitals, disabled persons' federations, and high-end elderly care institutions. Haier's W3, however, slashed the price straight into the 11,999 yuan bracket—on par with a mid-range electric wheelchair. That is an entirely different market.

Even more worth noting is the policy signal the host mentions:

"Domestic medical insurance can reimburse it, so that goes without saying."

If walking-assist exoskeletons make it into local medical insurance or long-term care insurance payment catalogs, it means their actual out-of-pocket price could drop further into the few-thousand-yuan range. That is decisive for the track: a breakout of consumer-grade products is never driven by manufacturers cutting prices on their own—it depends on payers turning "self-pay" into "partially reimbursed."

The host also did the math himself: a wheelchair costs money too, but no matter how cheap a wheelchair gets, navigating steps, climbing slopes, and having no one to push you remain real pain points; an exoskeleton, by contrast, preserves the inherent value of "a person can still walk on their own."

Real-World Scenarios: The Elderly, Rehab Patients, and That Lady from Argentina

The creator keeps steering the conversation back to a deeply personal motivation — getting his parents out and about:

"When I take my mom and dad out somewhere, they always find a spot to sit down while I wander around nearby... I don't know if, when they put this thing on, they could walk alongside me."

This description actually nails the single biggest real-world use case for walking assist exoskeletons: not getting paraplegics walking again (that's a long-term challenge for medical-grade rehabilitation exoskeletons plus neural interfaces), but extending the activity radius of older adults who can still walk but not very far — from "the neighborhood gate" to "touring scenic spots with their kids."

The most moving moment in the video comes from the Canton Fair, where an Argentine woman who had been in a wheelchair for years stands up after trying one on:

"Her friends were all in tears... She doesn't even know how many years she's been in a wheelchair, and she stood up and walked. And the more you use it, the smoother it gets."

Of course, a single anecdote can't replace clinical data — there's more than an order of magnitude of functional threshold between "being able to stand up and take a few steps" and "completing daily outings independently." But these on-site moments keep going viral because, for the first time, they shatter the stereotype that "exoskeleton = rehab device" in a way ordinary users can genuinely empathize with.

AI Adaptation: From "Wearable Device" to "Part of the Body"

The creator's description of AI, while colloquial, captures an industry consensus:

"It's got AI inside. It adjusts based on how fast you walk and your posture, so the more you walk, the smoother it feels — the more it becomes a part of your body."

Behind this lies the core tech stack that separates the consumer-grade exoskeleton from "passive support": gait phase recognition based on IMU + foot pressure sensors, adaptive joint torque control, and online learning tuned to individual differences. In other words, the same W3 should output different curves for an elderly user with a cadence of 90 steps per minute versus a young user at 120 steps per minute.

This also explains why "the more you use it, the smoother it gets" isn't just marketing speak — the moat of a consumer-grade exoskeleton lies not in the absolute value of motor torque, but in how quickly the algorithm converges to an individual's gait. On this front, domestic manufacturers are rapidly closing the gap with overseas leaders (brands such as Hypershell and Dnsys are also pushing a similar adaptive narrative).

Comparison with Unitree Robotics: General-Purpose Robots vs. Dedicated Walking-Assist Devices

In the latter half of the video, the creator turned his criticism toward Unitree Robotics (the transcript's is an ASR error) and its founder Wang Xingxing (is an ASR error), with a fairly emotional yet discussion-worthy argument:

"What Unitree builds for you is just that kind of robot, jumping and bouncing around — what does that have to do with me... Why don't you make something that actually benefits humanity?"

The creator's logic is: since China's robotics industry can already pull off backflips on two legs, why not build a walking-assist exoskeleton that "everyone can benefit from"?

This criticism doesn't really hold up, for three reasons:

1. Different market positioning: Unitree builds general-purpose humanoid and quadruped robot underlying platforms, emphasizing the transferability of the motion-control base; a walking-assist exoskeleton is a highly vertical special-purpose product, and the two reuse only partially overlapping tech stacks.
2. Different business models: General-purpose robots target the research, education, and secondary-development market; consumer-grade exoskeletons target C-end retail plus medical-insurance reimbursement — completely separate channel and after-sales systems.
3. "Backflips" aren't wasted effort: The creator himself later admitted those moves are pre-programmed, but being able to execute pre-programmed trajectories already means the joint torque, dynamic balance, and servo response have cleared the hardware threshold — which is exactly the hardware foundation for more complex walking-assist algorithms in the future.

Admittedly, "using the same hardware to both do backflips and provide walking assist" is not impossible in engineering terms, but reducing it to the founder's "skewed values" is more emotional expression from the creator than genuine industry analysis.

The Chinese Exoskeleton Track: Chengtian, Kenqing, Aosha, Hypershell, and Haier Share the Stage

In the video, the host mentions "there are other brands besides Haier," but doesn't name them. Based on the current state of the industry, the main players in China's consumer-grade / lightweight exoskeleton track can roughly be layered as follows:

• Chengtian Tech, Kenqing Tech, Aosha Intelligence, Yuanye Tech: Started in rehabilitation + industrial assist, with some products now reaching down into the consumer price band.
• Hypershell and Dnsys: Directly positioned for the outdoor and consumer markets, focusing on hiking/trekking assist, with a price range close to Haier's W3.
• Toread: An outdoor brand going cross-category, emphasizing mountain-climbing scenarios.
• Haier: A white-goods giant going cross-category; channel reach and brand recognition are its biggest advantages. The product itself follows a steady route of "lightweight + hip assist + AI adaptive."

The significance of Haier's entry isn't that its torque specs necessarily outclass those of Chengtian or Aosha — rather, it's that Haier has placed the exoskeleton into a retail setting for the first time: "stores in Shanghai, stores in Beijing, where you can walk in and try it on."

The maturation of a consumer-grade hardware category is rarely marked by a spec ceiling — it's marked by "try-it-on-in-store availability." That signals that channels, after-sales support, and user education have all reached their tipping point.

The host's judgment that "China's robotics are highly developed, so I'm fairly inclined to trust this thing" actually echoes an underlying industry pattern: the motor, reducer, and sensor supply chains for exoskeletons are heavily shared with those for quadruped and humanoid robots. China's accumulated production capacity in the latter is feeding back into lower BOM costs — and in turn, lower finished-product prices — for exoskeletons.

Expectations and Concerns: Beyond 1.8 Kilograms, What's Still Missing

On the optimistic side, products like the Haier W3 have genuinely pushed the consumer-grade exoskeleton to an inflection point where "ordinary families can start considering it." But as a review, several unavoidable questions remain:

• Is the 7 km endurance a real-world figure under full assist conditions? The gap between manufacturer claims and lab conditions, flat-ground conditions, and stair-climbing conditions is often 2-3x.
• Fall protection and self-balancing: The greatest safety risk of a walking assist exoskeleton isn't "insufficient assist," but "gait disruption caused by mistimed assist." The video does not address this part.
• Boundaries of applicable users: The creator mentioned the Argentine woman who "was in a wheelchair for years before standing up and walking." If a user has no voluntary muscle strength in the lower limbs at all, is hip assist alone enough to bear the full body weight? This involves strict delineation of medical indications.
• Battery life and charging: The most awkward scenario for a consumer-grade exoskeleton is "running out of power half an hour after going out." The real-world performance of endurance (in hours) requires long-term tracking.

Even so, the emotional landing point of the creator's line — "technology should benefit humanity" — is worth preserving. Not because it's sentimental, but because it points to a fact long obscured by the general-purpose robot narrative: what truly changes ordinary people's quality of life is often not the flashiest robot, but the most unassuming one.

If one day, when our generation grows old and can no longer walk more than a few steps, a 1.8-kilogram, AI-adaptive device partially reimbursed by medical insurance can accompany us for another ten years of walking — that would be the achievement most worth remembering from China's robotics industry.

Top Community Comments

@BuckwheatHusk01 (8 likes): Unitree builds general-purpose robots, not specialized ones — they're just on different paths. You can't pin this on Wang Xingxing's worldview; his mission may simply carry greater weight.

This comment is the most direct correction to the creator's "worldview" thesis — general-purpose robots and dedicated walking-assist devices are two parallel product lines. Moralizing what is essentially an industrial division of labor is, in itself, a logical misstep.

@qwei312 (7 likes): This has nothing to do with worldview, does it?

A more concise version. Choosing a technical roadmap is a business judgment, not a moral one — this comment captures the instinct of most reasonable viewers.

@WendaoPingzhang (6 likes): Unitree builds the underlying architecture — don't get that wrong. For applications, you just buy it back and train it yourself, like training on your own big data.

This comment touches the very core of the industry: general-purpose robots provide a "motion-control base + secondary development interface," and specific applications (including walking assist) can be built on top of it via transfer learning. This is precisely why the creator's question — "why not just build a walking-assist device directly?" — conflates the platform layer with the application layer.

One-Sentence Summary

The Haier W3 has simultaneously pushed the price, weight, and accessibility of consumer-grade walking assist exoskeletons to a new tipping point (1.8 kg / 11,999 RMB / in-store try-on available), making it worth recommending to families with elderly members who "can walk but not far"; however, the 7 km battery life, fall protection, and the boundary of medical indications still require more third-party real-world testing and clinical data to validate — and drawing a "core values" comparison between it and Unitree's general-purpose robots makes for an interesting but not-quite-valid industry analogy.