"Will it fall?"
This is the thought that flashes through the mind of everyone placing a window-cleaning robot onto glass for the first time. No manufacturer dares write "never falls" in its marketing; when work at height is involved, no one can bear the responsibility behind those four words.
What a window-cleaning robot sells is not just cleaning power, but the promise that risk has been taken over by technology.
Window-cleaning robots are becoming an increasingly visible variable in the home robot industry.
IDC data shows that in the second quarter of 2026, global shipments of window-cleaning robots reached about 847,000 units, up 59.8% year-on-year, with growth significantly higher than the 21.5% growth of the overall household cleaning robot market over the same period. AVC data shows that from 2020 to 2025, the Chinese market grew from 480 million yuan to 1.7 billion yuan, a compound annual growth of 28.6%.
Capacity is also expanding: according to Cailian Press, Ecovacs' Winbot in its early years produced only a few thousand to ten- or twenty-thousand units per month, and after ramping up in 2025, monthly capacity jumped from the 100,000-unit level to the 200,000-unit level. In 2025, global window-cleaning robot shipments were about 2.373 million units, up 70.4% year-on-year, and the industry expects this to exceed 3 million units in 2026.
Behind the volume growth is, essentially, the growth of "trust".
01 A "small category" that is scaling up
Cleaning windows is probably the household chore one should least do by hand: infrequent yet essential, simple yet dangerous.
High-rise apartments, large floor-to-ceiling windows, and sealed exterior windows have long been cleaning dead zones. There are only two traditional options: lean out and take the risk, or pay someone to do it.
Robot vacuums solve tasks you can do but don't want to; window-cleaning robots solve tasks that are very hard to do safely by yourself.
The difference lies in the consequences. A robot vacuum missing a corner is only an experience issue; a window-cleaning robot misjudging a window edge is a safety issue. So its first value is not saving time, but removing people from danger.
But the other side of market expansion is that unstable user experience gets amplified at the same time. With monthly capacity jumping from ten or twenty thousand units to the 200,000-unit level, any fluctuation in a single component will be magnified by shipments in the millions.
Consumers want the machine to take the risk for them, yet in actual use they still need to attach the safety rope, check the glass conditions, and supervise operation. Long-term testing by overseas review organizations shows that such products can handle dust and daily maintenance, but have limited ability against stubborn stains, and in most scenarios still depend on manual placement, transfer, and supervision.
This is not a new problem. Corners left uncleaned, loud noise, and suitability only for large, flat glass were the norm for early window-cleaning products; after several years, the basic experience has improved dramatically, but edge blind spots, stubborn stains that won't come off, and the need for human supervision remain shortcomings common to all brands' products.
For a machine working against glass dozens of stories high, a millimeter-scale blind spot corresponds to a real difference in how "controllable" the machine feels to the user.
Li Xiaohua, product line director of Ecovacs Winbot, described an internal debate at a communication meeting: in the early years, some on the team believed that if a particular window occasionally got stuck and 95% of scenarios were covered, it "should basically satisfy user needs".
But in-home research overturned that idea. One user had a large floor-to-ceiling window where the machine got stuck in the far corner every time it cleaned; although the stuck point was on only that one window in his home, it got stuck every time, and the user's choice was simply not to clean that window at all.
"If that window is the main window in his home, then for him it's not a 1% stuck-rate problem," Li Xiaohua said. "It's a 0-and-1 problem."
Add to that the safety function window-cleaning robots bear, and consumers' tolerance for them is naturally even lower.
What's harder is mass production.
Building one good prototype doesn't mean you can reliably build millions. A window-cleaning robot's safe adhesion depends on the fan, seals, air-replenishment efficiency, structural strength, and sensor coordination. The upper and lower limits of the fan's air-replenishment efficiency, the rebound consistency of seal foam after aging — these variables hidden in the parts can all affect the stability of the whole machine on the window surface. Competition will ultimately enter the manufacturing stage.
Ecovacs says its R&D and manufacturing use a coordinated mechanism: when structures, components, or functions change, processes, tooling, and inspection methods are adjusted in sync, and products must pass multi-dimensional verification covering function, performance, safety, and abnormal operating conditions. What it must solve is turning laboratory metrics into production-line quality control, and then into a stable experience in the hands of consumers.
Standards set the industry's common floor; corporate standards answer whether a product can keep working in more complex environments, and whether that capability can be replicated in every single unit.
The gap between "meeting the standard" and "leading in experience" lies precisely here.
02 The first industry standard arrives, but "meeting the standard" is not the finish line
On October 13, 2026, the first industry standard for window-cleaning robots, QB/T 8242—2026 "Household and Similar Window-Cleaning Robots", was publicly promoted, taking effect on December 1. The standard was led by the China Household Electric Appliance Research Institute, with Ecovacs among the participating whole-machine enterprises, alongside Shanxi Jiashida, Beijing Hter, and several testing institutions.
The standard covers operational protection, cleaning performance, test methods, inspection rules, marking, and instructions for use, defines scenarios such as vertical flat glass, glass with adhesive-edged frames, glass without adhesive-edged frames, and frameless glass, and defines concepts such as safety device, cleanliness, and cleaning cycle.
What it targets is a long-standing difficulty in comparison. Brands explain adhesion and cleaning capability in different ways, making it hard for consumers to understand the parameters. The standard translates brands' language into unified test methods, turning "trust" into measurable indicators.
The first version of the industry standard set quantitative thresholds: overall cleanliness of no less than 70%, edge-and-corner cleanliness of no less than 65%, and operational durability of no less than 200 hours.
But standards have boundaries. They must be strict enough, yet not so high that most companies cannot reach them. An industry standard is a floor, not a ceiling.
For the window-cleaning robot category, the first challenge the industry standard faced was not even setting metrics, but making definitions: what counts as adhesion failure or abnormal operation, and how to classify conditions like power loss, getting stuck, and obstacle crossing. Before 2026, every brand had its own interpretation of these concepts. The first thing the first version of the standard did was establish unified definitions for these terms.
Li Xiaohua also revealed another trade-off in the drafting process. Ecovacs' internal testing standards cover far more scenarios than the industry standard. "If we applied Ecovacs' current testing standards and included all user scenarios as test items, we worried that when the industry standard was first introduced, not many products in the industry would truly be able to meet it."
In other words, the "low bar" of the first industry standard was partly the result of deliberate trade-offs. First establish the standard, then talk about raising it. Under the current mechanism, a standard cannot be adjusted for two to three years after publication, and a revision takes about a year from initiation to implementation, meaning this system will remain stable for three to four years. The industry standard set no high thresholds for performance indicators such as cleaning effectiveness, accomplishing the journey "from 0 to 1."
In the view of Lei Feng Network (WeChat account: Lei Feng Network), a question follows: the standard can define baseline safety and cleaning capability, but it is difficult for it to cover all household windows.
Glass under standard test conditions is only the starting point; irregular windows, narrow windows, and window surfaces with obstacles are the real dividing line. If a "compliant" machine gets stuck, misses spots, or leaves water streaks in a real home—something that often requires human intervention—how should the word "compliant" be understood?
This is precisely the tension between industry standards and corporate standards, and also the divide between leading players and ordinary players.
03 Ecovacs' Real Test
Ecovacs was one of the earliest entrants into this category. R&D began in 2011, and its first product, the Window Bot, launched in 2013.
By entering early, it also gained an additional layer of advantage.
Ecovacs had made the floor-cleaning robot (Floor Bot) for years: the bumpers and ball-joint detection on the front of the Window Bot came from the migration of the Floor Bot design, as did its cloth-based cleaning method; meanwhile, the engineering know-how the Window Bot developed in vertical-surface adhesion and edge detection accumulated into the company's overall sensor and path-planning capabilities.
Ecovacs summarizes this as the "flow of foundational technologies" between different product lines.
This explains why the Window Bot iterated faster than the category itself, and it also points to its next step: the perception upgrades and cleaning-structure evolution the Floor Bot went through will most likely replay on the Window Bot.
Its product roadmap has roughly gone through three stages:
standalone machines solving the problem of staying attached;
base stations enabling a complete cleaning cycle;
and automatic mop cleaning to further reduce user intervention.
The base station products launched between late 2023 and 2024 combined the power supply and safety rope into one, with a built-in battery as a backup for low-power situations, which Ecovacs regards as a turning point.
The market's response was direct.
According to AVC (All View Cloud) data, in 2025 Ecovacs accounted for over 60% of domestic window-cleaning robot sales in China, ranking first in the Chinese market for five consecutive years, with a 62.3% share in the first quarter of that year; in the first quarter of 2026, it held a 96.5% share of the online high-end market above 2,000 yuan. According to IDC data, in 2025 it shipped about 1,195,000 units globally with a 50.30% share, and in the first half of 2026 shipped 1,075,000 units with a 57.60% share, ranking first worldwide.
But being the market-share leader also means bearing a heavier burden of explanation. A company selling "safety assurance" must accept stricter scrutiny than one selling "cleaning efficiency."
In the competition among window-cleaning robots, the ultimate question is not who makes their machine look more like a robot, but who can compress uncertainty the furthest. Competition has already extended from features to standards, supply chains, manufacturing, and global service systems.
In its next stage, the industry must solve at least three things: adaptation to complex window types, the upper limit of cleaning effectiveness, and further reduction of human intervention.
Modern households hope to hand dangerous, tedious, low-frequency chores over to machines. But technology has not eliminated risk; it has merely shifted risk into product design, standard testing, user operation, and corporate responsibility.
A standard is the starting point of trust, but it is not trust itself.
04 Conclusion
Ecovacs believes that the future household robot will not necessarily be an "all-purpose robot" that does everything, but more likely a collaboration between general-purpose and specialized robots:
embodied intelligence will be responsible for understanding tasks and dispatching equipment, while what actually climbs onto exterior windows will still be a machine specifically designed for glass, frames, air pressure, and the risk of falling.
The significance of the Ecovacs case lies not only in its market-share leadership, but in the fact that it witnessed and participated throughout the formation of an emerging category.
From starting R&D in 2011 and launching its first product in 2013, from standalone machines to base stations to participating in drafting the first industry standard, and finally expanding overseas—what Ecovacs has traveled is both a product's life cycle and a typical way for a Chinese company to define an emerging category.
But exporting rules is harder than exporting products.
During the group interview, Li Xiaohua admitted, "If global standards are ultimately defined by two or three domestic brands themselves, whether they can actually exert pull on other overseas brands, countries, and users is a fairly major point."
In Leiphone's view, the various figures cited in the article use inconsistent statistical criteria, but they all point to one thing: this category has entered a phase of large-scale growth, and Ecovacs now stands in a position where more and more people are scrutinizing it through a magnifying glass.
Once the standard is implemented, it will attract more players to enter the market, and product prices, functions, and forms will all become richer. Entering the category first does not mean retaining the right to define it—ultimately, the market belongs to those who turn certainty into industry common sense.
This is an original article by Leiphone. Unauthorized reproduction is prohibited. For details, seeReprint Notice。