The hardest part of building a home robot isn’t AI. It’s earning our trust

Estimated reading time: 18–20 minutes

Editorial disclosure: Ecovacs Robotics invited Digital Reviews Network, along with ten other media attendees, to Suzhou, China, and covered flights, accommodation, meals and local transport for this media visit. In accordance with DRN’s editorial guidelines, Ecovacs had no editorial input into this article and did not review it before publication.

There is something inherently theatrical about a robot.

Place one in a room and ask it to collect the shoes strewn across the floor. Ask it to tidy a workspace and return the pens to their holder. Have it pick up dirty laundry and place it into a washing machine.

We are naturally drawn to the moment when a machine appears to understand the world around it.

That spectacle — the intelligent robot perceiving its surroundings and acting with purpose — has shaped much of the public conversation around home robotics.

We talk about artificial intelligence, computer vision, object recognition and autonomous navigation. We compare suction figures, mapping systems, camera resolutions and the increasingly improbable number of household chores manufacturers believe can be automated.

Consumers rarely buy a robot because it uses SLAM, neural networks or computer vision. They buy it because they don’t want to vacuum, mow the lawn, clean the windows or skim the pool. The technology only matters if it quietly removes one more repetitive task from everyday life.

Before travelling to China to meet the Ecovacs Robotics team, I expected to spend a great deal of time discussing exactly those things.

We certainly had plenty of those conversations.

But after less than two days of product demonstrations, roundtable discussions, laboratory tours and production-line briefings, the message that stayed with me was not necessarily about the technology designed to attract attention.

It was about the technology designed to prevent anything interesting from happening at all.

A motor that does not overheat.

A wheel assembly that continues operating after years of movement.

A battery cell with a consistently coated electrode.

A robot that avoids scratching a floor, becoming stranded under furniture or waking a household unnecessarily.

None of this makes for a dramatic product demonstration.

Yet it may represent the most important work being done in home robotics today.

The challenge is no longer simply making a robot capable of performing a task. It is making that robot reliable enough, predictable enough and unobtrusive enough that people are willing to hand the task over permanently.

That trust extends well beyond whether a robot finishes cleaning. It includes safety, reliability, privacy, long-term support and confidence that the machine will continue doing its job long after the novelty has worn off.

The real challenge is earning our trust.

Consumers buy outcomes. Engineers chase edge cases.

Ecovacs began life in 1998 as an original equipment manufacturer before launching its own brand in 2006.

Today, Ecovacs’ ambitions extend well beyond the robot vacuums that first made its name, reaching homes in more than 180 markets.

Rather than viewing vacuuming, window cleaning, lawn maintenance and pool care as separate product categories, the company increasingly sees the home as a single living space made up of repetitive tasks waiting to be automated.

That philosophy is reflected in an ecosystem spanning the Deebot range of floor-care robots, Winbot window cleaners, and more recently in Australia and New Zealand, the GOAT robotic lawnmower and Ultramarine pool-cleaning robot.

From the outside, that expansion looks like a company placing robots wherever it can find an unpleasant household chore. From the inside, the process is more complicated.

Every new environment introduces a new collection of variables, failure points and uncooperative real-world conditions.

A robot vacuum may need to identify shoes, charging cables, toys, chair legs and the countless objects a family casually leaves scattered across the floor.

Most of us would barely notice the clutter. The more conscientious amongst us would tidy up without a second thought. A robot has to recognise it, classify it and decide what to do next.

A window-cleaning robot must remain attached to a vertical pane while contending with grime, bird droppings, weather exposure and, in coastal Australia, salt mist.

It also has perhaps the highest stakes. If a robot vacuum makes a mistake, it gets stuck. If a window robot loses grip on a second-storey window, the consequences become far more serious.

In contrast, a lawnmower must deal with changing weather, uneven ground, overgrown grass, temporary obstacles and the fact that an outdoor environment can look different almost every day.

Similarly, a pool robot faces water, leaves, dust, curved surfaces, stairs, corners and pools that do not always conform to the neat rectangular shapes shown in marketing material.

The more time I spent listening to Ecovacs’ engineers and product teams, the clearer it became that the headline feature is often the easy part. The difficult part is everything around it.

What surprised me most wasn’t the machinery. It was the people and company culture. Almost every conversation inevitably drifted away from marketing features and back towards edge cases: door thresholds, pet hair, Australian grasses, salt spray, battery longevity and the tiny failures that most consumers would never know existed.

Those conversations weren’t driven by marketing wish lists. They were informed by home visits, weekly customer telemetry, app questionnaires and the often brutally honest feedback gathered from Reddit, social media and customer communities.

It was a reminder that while consumers compare feature lists, engineers spend their days obsessing over exceptions and “what ifs”.

Walking through the central laboratory felt less like touring a consumer electronics company and more like visiting an engineering research facility.

Inside Ecovacs’ electromagnetic compatibility testing chamber, products are tested both for the interference they generate and their ability to continue operating under external electromagnetic disturbance

As an engineer at heart, it was impossible not to appreciate the purpose-built test rigs lining the laboratories, where robots quietly repeated the same movements thousands of times under the watchful eye of cameras, sensors and measurement equipment.

Cleaning performance is evaluated using standardised dust, particles and hair across both hard floors and carpet. Ecovacs has developed its own straight-line and box-based test systems to improve repeatability, with testing covering both new products and performance validation before over-the-air software updates are released.

Having reviewed both the Deebot T50 Max Pro and, more recently, the flagship Deebot X8 Max Pro, it was fascinating to see the controlled testing environments that underpin their performance claims. Watching engineers validate navigation, obstacle avoidance and cleaning performance made it clear that what appears effortless inside a customer’s home is built upon thousands of carefully engineered repetitions inside the laboratory.

Controlled test environments allow cleaning performance to be measured consistently across different flooring materials and debris types.

There are heat tests to ensure that components such as motors and batteries remain within prescribed surface-temperature limits. Electromagnetic compatibility testing measures the emissions generated by a product and then subjects it to external interference to determine whether it can continue functioning correctly.

Noise is captured using 10 microphones positioned around the robot, providing engineers with data they can use to identify and reduce unwanted sounds.

Then there is longevity.

A Deebot undergoing laboratory testing. Much of the work behind a dependable robot is designed to remain invisible to its eventual owner.

Every home robot contains dozens of small, constantly moving components.

Each of these creates another opportunity for wear. Turntables, rollers, drive wheels and side brushes must all survive repeated movement, impacts and changes in direction.

Before a robot ever reaches someone’s home, it must survive the journey there. Ecovacs uses automated vibration testing to simulate the repeated movement, impacts and stresses experienced during shipping, helping identify packaging and durability issues before products leave the factory.

One of the more memorable descriptions from the laboratory tour was that Ecovacs uses robots to test robots. Mechanical rigs repeatedly operate the moving parts of a product to expose weaknesses that might otherwise take years to appear in ordinary use.

It is an oddly poetic image for the industry: robots spending their working lives testing other robots so that future owners never have to think about them.

Automation, quietly improving automation.

This is the work consumers rarely get the opportunity to see.

As consumers, we notice when a wheel becomes stuck.

We do not see the engineers attempting to reproduce the exact combination of surface, angle and obstruction that caused it.

When a robot fails, we rarely think about the millions of successful cleaning cycles performed by every other machine. We remember the failure, tell our friends about it and sometimes loudly decide that the brand cannot be trusted.

Good engineering is often invisible. Perhaps the greatest compliment a home robot can receive is that its owner eventually stops thinking about it altogether.

Local problems require local solutions

One question lingered in the back of my mind before the trip: were Australia and New Zealand genuinely influencing product development, or were they simply smaller extensions of a much larger global market?

Then there was Patrick Pilcher, better known as the Witchdoctor NZ. After countless conversations covering everything from robot vacuums to novel writing, this imported Aussie realised just how much there was still to learn.

Australia and New Zealand might often be bundled together, but they are far from the same market.

There is always a risk that products designed around homes in China, Europe or North America will be shipped into Australia with little more than a new power plug and a localised app.

Or in the case of New Zealand, an Australian app.

The conversations suggested something more considered. Rather than asking Australia and New Zealand to adapt to global products, much of the discussion centered on adapting products to local homes.

What impressed me most wasn’t simply being given access to Ecovacs’ product team. It was the openness of the discussion. At no point did anyone skirt a difficult question or attempt to steer the conversation back towards marketing talking points.

The ANZ media roundtable brought Ecovacs engineers, product managers and regional leadership together with Australian and New Zealand media to discuss real-world product challenges and future development.

Just as importantly, the discussion wasn’t a one-way interrogation of the Ecovacs team. Questions flowed in both directions. The engineers and product managers were just as interested in understanding where their products struggled in Australian homes as they were in explaining how they worked.

The media didn’t replace Ecovacs’ own research efforts. Instead, our experiences added another layer of context to the home visits, customer telemetry and questionnaires the company was already using to guide development.

There was remarkably little discussion about beating competitors. The conversation focused on homes: the awkward threshold that catches a wheel, the pool with an unusual shape, the dog that sheds year-round, the lawn that refuses to behave like the one in the product photos.

The focus wasn’t the robot. It was the environment the robot had to perform in.

That perspective challenges almost every engineering decision. A threshold isn’t simply a raised strip between rooms. It’s a failed cleaning cycle waiting to happen.

Australian homes present a unique challenge. We build larger homes than many parts of the world, carpet remains common, our blocks are often bigger, our lawns are harsher, and our climate can swing from dry inland heat to salt-laden coastal air.

Designing for Australia isn’t simply a matter of translating the app into English. It means understanding how Australians actually live.

Even our enthusiasm for large windows, sliding doors, glass pool fencing and open-plan living creates a distinct set of requirements.

Ecovacs said its Australia and New Zealand work has included higher-airflow motors for carpet-heavy homes, mower development informed by buffalo and kikuyu grass, and consideration of Winbot use on large expanses of glass.

That does not mean every local problem has been solved.

The existing window-cleaning technology was described during the discussions as useful for maintenance, but not necessarily capable of applying the pressure required for difficult exterior grime or bird droppings. Salt mist creates another complication, particularly near the coast.

The most useful discussion here wasn’t about cleaning performance at all. It was about safety.

Ecovacs explained that the engineering philosophy begins with preventing a fall rather than insuring against one. Suction performance, edge detection, environmental sensing, onboard battery backup and the Omni Station all form successive layers of protection designed to ensure the robot never loses contact with the glass. Insurance exists as a final safeguard, but the objective is that it should never be needed.

Digital Reviews Network has reviewed both the Winbot W1 and more recent Winbot W1S, seeing the testing behind the products provided valuable context for how Ecovacs has approached safety and reliability over successive generations of the platform.

Winbot undergoing repeated testing on a dedicated glass rig inside Ecovacs’ central laboratory. The company’s approach is to validate multiple layers of protection so the robot remains securely attached long before insurance would ever become necessary.

Pool cleaning brings its own unresolved challenges.

One takeaway from the Ultramarine discussions particularly stayed with me. Unlike a floor, the environment inside a pool is never truly static.

Leaves float. Fine dust settles. Stir up the silt on the bottom too aggressively and the robot may simply suspend the debris back into the water column, forcing it to wait before cleaning can continue. Intelligence alone doesn’t solve that problem. Sometimes patience is the most efficient cleaning strategy.

Reliability is a significant concern in the Australian market, where some products have developed a reputation for failing within a relatively short period. Large pools, unusual shapes, trees, dust and beach-entry designs can expose weaknesses that are less obvious in controlled demonstrations.

Shallow sections are particularly difficult. A cleaner may physically reach the area but no longer have sufficient water above it to maintain effective intake.

These are not problems that can be solved by adding “AI-powered” to a product description.

They require mechanical design, fluid dynamics, materials testing, software refinement and, frequently, the humility to accept that the first attempt will not be perfect.

One phrase that came up repeatedly during the discussions was “honest engineering“. The idea was refreshingly simple: don’t add technology for the sake of a feature list. Add it because it genuinely solves a problem customers face.

Ecovacs described its first pool-cleaning generation as intentionally simple: a foundation from which the company can collect feedback and build a more mature product.

That is arguably a less exciting message than promising a revolution.

For a company trying to earn consumers’ trust, acknowledging that not every problem has been solved may be the most encouraging message of all.

Artificial intelligence meets the real world

The limits of current generation robotics became particularly clear during discussions about GOAT robotic lawnmowers.

The GOAT project began with an assumption: much of the technology developed for Deebot and Winbot could simply move outdoors.

It didn’t. In fact, it couldn’t. It proved to be a costly assumption.

According to Ecovacs, only around 10% of that technology ultimately proved reusable. It was a reminder that artificial intelligence doesn’t replace engineering. It builds upon it.

Moving from the controlled environment of a home to the unpredictability of a garden wasn’t an evolution of existing robotics. This wasn’t simply a case of making a robot vacuum waterproof and fitting it with blades.

It required rethinking almost everything.

Indoors, the robot operates within a comparatively constrained space. Walls generally remain where they were yesterday. Furniture may move, but the overall structure of the environment is recognisable.

Outdoors, almost everything can change. Daily.

Multiple GOAT robotic mowers operating in an outdoor test environment—a far less predictable setting than the relatively stable interior of a home.

Grass grows. Rain alters the surface. Branches fall. Pets dig. Children leave toys behind. Cars are parked across previously accessible paths. Shadows move throughout the day, and the same patch of ground can look markedly different depending on the season.

One of the first engineering challenges was removing one of the biggest barriers to robotic mower adoption: installation.

When Ecovacs began developing GOAT around four years ago, one of the first major engineering breakthroughs was eliminating the need for traditional boundary wires through wireless positioning and vision-based navigation. It solved one of the biggest barriers to robotic mower adoption, but it didn’t eliminate every navigation problem.

During the roundtable, one of the media attendees described the Ecovacs GOAT O600 RTK mower repeatedly becoming stuck in the same uneven areas and struggling when a temporary object, such as a parked car, appeared along a previously mapped route.

What happened next surprised me. Rather than defending the product, Ecovacs openly acknowledged that GOAT doesn’t yet possess the same dynamic learning capability as its indoor Deebot products.

The outdoor environment is substantially more complex, and for now the practical solution is often surprisingly simple: manually define no-go zones where persistent obstacles exist.

It was an unusually useful admission.

Technology companies are often more comfortable discussing what their products can do than where their limitations remain.

Those limitations became some of the most interesting parts of the discussion. Understanding them is essential if robotics is to mature.

Artificial intelligence can improve object recognition, navigation and decision-making, but it does not rewrite the laws of physics.

A robot still needs traction.

It still needs sufficient battery capacity.

It needs a chassis capable of crossing uneven surfaces and hardware that can tolerate dirt, moisture, heat and repeated impacts.

The most sophisticated algorithm in the world cannot compensate indefinitely for a wheel that cannot climb the terrain in front of it.

The industry’s fascination with AI can sometimes obscure that fact.

Intelligence matters, but only when it is embodied in a machine capable of performing reliably in the messy, unpredictable environments we call home.

Of course, we couldn’t resist asking whether “GOAT” was a nod to “Greatest Of All Time“.

The answer was much simpler.

Goats eat grass.

The battery factory changes the scale

Where the robotics laboratory demonstrated how Ecovacs tests ideas and products, the battery factory demonstrated how those ideas become products.

Walking into the lithium-ion battery factory was a lesson in precision and patience: the physical foundation upon which all of Ecovacs’ software and engineering ultimately depends.

Battery cells are among the least visible components inside a consumer product. Most people will never see one outside its enclosure, yet battery quality influences runtime, lifespan, charging performance, safety and ultimately whether the product remains useful several years after purchase.

Ecovacs operates its own battery manufacturing facility in Nanxun, about ninety minutes by road southwest of Suzhou.

The production line stretched from raw electrode preparation through to final inspection and packaging, with automation handling the process from powder mixing and slurry coating through pressing, winding, electrolyte injection, formation and ageing.

Electrode material is applied to metal foil before passing through multiple quality-control stages designed to detect even microscopic inconsistencies. Every subsequent step, from pressing and winding through electrolyte filling and sealing, is tightly controlled because tiny manufacturing variations can eventually become battery failures.

Electrode material being processed onto metal foil. Small inconsistencies introduced here can become capacity, reliability or safety problems much later.

What struck me was not simply the automation, but the patience built into the process.

In consumer technology we’re conditioned to think faster is always better. Faster processors. Faster charging. Faster product cycles.

Battery manufacturing works to an old-world timetable.

After the electrolyte is injected, the cells are left for approximately 48 hours to allow it to penetrate the electrode structure. The first low-rate charge forms the solid electrolyte interphase, a protective layer created through the cell’s own electrochemical reaction.

Gas produced during this initial charge is captured, removed under vacuum and the cell is resealed. The cells then undergo three days of high-temperature ageing, followed by another two to three days at room temperature before grading and final inspection.

In an industry obsessed with speed, some things cannot simply be rushed.

The cells must be allowed to settle. Defects must be given time to emerge. Electrical characteristics must be measured across the batch, and the physical packaging inspected for surface or film damage.

Ecovacs’ lithium-ion battery production facility in Nanxun, where automated processes extend from electrode preparation through cell assembly, formation, ageing and final inspection.

Much of modern consumer technology is designed to feel instantaneous.

Tap the screen. Start the robot. Ask the assistant. Receive the answer.

The factory was a reminder that the physical systems supporting those experiences are built through slower and less glamorous processes.

Precision matters because small inconsistencies are multiplied when thousands of cells move through a line.

A coating defect measured at the electrode stage may become a capacity, reliability or safety problem much later.

Again, the most impressive outcome was the complete absence of drama. No sparks. No smoke. No excitement. Just thousands of batteries quietly progressing through a process designed to ensure that years later, somebody presses a power button and never gives that battery a second thought.

The cell works.

It continues working.

Nothing overheats, swells or fails unexpectedly.

Just like the best robotics, the battery disappears into the background.

Consumers see battery life. Engineers see coating thickness, formation cycles and microscopic defects that determine whether that battery will still be working years later.

We asked about emerging battery technologies, including silicon-carbon anode. Ecovacs’ response was telling. Rather than chasing the next chemistry, the company has invested heavily in refining its existing lithium-ion manufacturing process. That same factory doesn’t simply produce batteries for Ecovacs’ own robots, it also manufactures battery packs for other companies.

This ownership of the supply chain gives Ecovacs a level of vertical integration that’s unusual in consumer robotics.

Owning the battery production line also gives Ecovacs greater control over quality, supply and cost. But walking through the factory, I came away with a different impression.

The battery wasn’t simply another component.

It was another example of the same philosophy I’d seen throughout the trip.

Reliability isn’t something that happens by accident.

It’s engineered into the product, one carefully controlled process at a time.

Automation dominates much of the production line, but human inspection and assembly remain part of the wider quality-control process.

Trust extends beyond navigation

Until this point, most of what I’d seen could be measured: temperatures, noise, durability, battery quality and cleaning performance.

Trust introduced a different challenge. Some of the most important questions surrounding modern robotics cannot be answered with a laboratory instrument.

The engineering problems become human ones. As robots gain cameras, detailed maps and greater autonomy, the challenge is no longer simply what they can clean, but what they can observe, remember and communicate.

The most important question becomes what happens to the data they collect.

A serious security vulnerability affecting the Deebot X2 series was publicly disclosed in 2024. The company’s response at the time attracted criticism from some observers, highlighting how quickly trust can be lost when connected products are involved.

Ecovacs now says “security starts when data is generated“. Assurances are important, but consumer scrutiny will only increase.

A floor map may appear harmless until it reveals the layout of a home.

A camera may help a robot avoid a pet, but it also creates obvious questions about image processing, cloud access and retention.

A security-patrol feature may be useful, yet it moves the robot closer to becoming a mobile surveillance device.

The more capable the robot becomes, the more clearly companies will need to explain what information is captured, where it is processed and how long it is retained.

Privacy is only one dimension of trust. Longevity is another.

Consumers see a robot vacuum. Building one increasingly requires expertise that extends beyond robotics into software engineering, cybersecurity and privacy.

Ecovacs said its products are designed around a lifespan of approximately six to eight years, although owners may replace them earlier as their household requirements change. It also intends to retain spare parts for at least five years after a product reaches the end of its commercial life.

Self-repair kits containing parts, tools and instructions are being trialled for markets including Australia and Europe, although no firm Australian timeline was provided.

Repairability may become one of the most important tests of the industry’s maturity.

For that trust to endure, a home robot must remain repairable, supported and useful for a reasonable period.

Like much of the consumer electronics industry, robotics has conditioned buyers to expect rapid product cycles. That pace may drive innovation, but it can also create anxiety that an expensive product will quickly become obsolete or unsupported. Long-term trust depends on knowing it will remain supported, maintained and repairable well beyond launch.

The roundtable discussions reinforced that trust doesn’t end when a robot leaves the factory. Servicing, software updates and customer feedback were presented as part of a continuous engineering loop, with issues encountered in customers’ homes feeding back into future product development.

That philosophy mirrored what I’d already seen in the laboratories, where even over-the-air firmware updates are validated through structured testing before release. Reliability isn’t measured only by how a product performs on launch day, but by how well the manufacturer supports it throughout its life.

The same thinking also shapes product strategy.

Flagship models are expected to showcase the company’s latest innovations, with new generations typically arriving every eight to ten months.

Entry-level and mid-range products evolve more deliberately, prioritising stability and longevity over rapid feature additions.

Not every customer is looking for the newest technology. Many simply want a robot they can rely on for years.

Software updates, replacement components and realistic repair options will ultimately do more to build trust than another increase in the advertised suction number.

From individual appliances to a robotic home

Ecovacs’ longer-term ambition is not limited to building a better robot vacuum.

Over breakfast, Karen Powell, Regional Director of Ecovacs Robotics Australia and New Zealand, returned to a phrase that had quietly underpinned everything I’d seen during the trip: “Robotics for All.”

During my time with Ecovacs, that same philosophy appeared repeatedly.

Whether I was speaking with engineers, product managers or regional leadership, there was remarkably little discussion about building individual appliances. The conversation was almost always about building a more capable home.

The company increasingly speaks about a home robotics ecosystem: different machines responsible for different surfaces and environments, potentially sharing data and working in collaboration within the wider home.

The individual robots matter, but after two days in Suzhou I came away with the impression that Ecovacs no longer thinks in terms of products. It thinks in terms of robotic capabilities that can be applied across different environments and entirely new categories.

That ambition is already visible in its expansion from floors to windows, lawns and pools.

The philosophy isn’t limited to future concepts. It’s already beginning to shape products available today.

It is also evident in the company’s partnership with Bosch, which combines Ecovacs’ robotic technology with Bosch’s experience in integrated appliances. The result is a robotic vacuum system designed to disappear into the home itself rather than occupying a conspicuous dock against the wall. Bosch holds sales rights for the collaboration in Europe and Australia.

The idea is significant.

For years, household robots have felt like technology added to the home rather than designed as part of it.

The next phase may involve designing them as part of the home from the beginning.

A robot could emerge from cabinetry, perform its task and return without requiring the owner to accommodate a large base station in the middle of a room.

That broader ambition is also visible elsewhere within the group. Tineco’s Shiwan Tiangong embodied cooking system links AI-generated recipes with automated cooking and self-cleaning, extending the idea of household robotics beyond cleaning surfaces and into preparing meals.

Shiwan Tiangong isn’t intended as tomorrow’s kitchen appliance. It’s a glimpse of Ecovacs’ longer-term ambition: moving from individual robots that perform single tasks towards coordinated systems capable of assisting throughout the home.

Until then, I’d largely thought about household robots as individual appliances. Shiwan Tiangong challenged that assumption by treating cooking as a complete workflow rather than a single task.

It is a significant shift. A robot vacuum automates one repetitive task. Shiwan Tiangong attempts to coordinate software, sensing, physical movement and cleanup across an entire domestic process.

The significance is not that every kitchen will soon contain one. It is that the category is beginning to move from isolated appliances towards systems capable of understanding and completing more complex household routines.

At the beginning of this article, I asked you to imagine a robot picking shoes up off the floor, returning pens to a holder and loading dirty clothes into a washing machine.

That robot already has a name. Ecovacs calls it Bajie.

It isn’t a product you can buy today. It’s a glimpse of where the company believes home robotics is heading. Seeing several Bajie prototypes in action made the opening thought experiment suddenly feel much less like science fiction.

Bajie isn’t alone. Another concept, Maotuan’er, pointed in a very different direction. Rather than automating another household chore, the companion concept explored how robotics might become more expressive, responsive and emotionally legible inside the home. It suggested Ecovacs’ vision is not only about what robots can do for people, but how naturally people might live alongside them.

It was perhaps the clearest indication that Ecovacs’ long-term ambitions extend beyond automation alone. If Bajie represents the practical future of household assistance, Maotuan’er hints at something more human — a future where robots are designed not only to work alongside us, but to interact with us in more natural and engaging ways.

Ideas may become future features in existing product lines, or new product categories. Others may never progress beyond the discussion stage.

That uncertainty matters.

Throughout the discussions there was a recurring sense of what Ecovacs called “honest engineering”. Not every idea deserves to become a product, and not every feature deserves to exist simply because it is technically possible. The measure was always whether it solved a real problem for the customer.

A vision should not be mistaken for a roadmap, and an interesting engineering suggestion is not the same thing as a commitment to deliver it.

However, taken together, the conversations reveal how Ecovacs increasingly views the home.

Not as a collection of isolated chores, but as an environment filled with surfaces, objects and repetitive tasks that can gradually be mapped, understood and automated.

The risk is that every manufacturer attempts to build a closed ecosystem.

A truly useful robotic home will eventually require some level of interoperability. Consumers should not need separate islands of automation that refuse to communicate simply because the mower, vacuum, security system and energy platform carry different logos.

That may prove to be a harder challenge than building any individual robot.

After spending two days inside Ecovacs, I no longer think the company’s biggest challenge is building a robot capable of recognising a shoe or mowing a lawn.

The real challenge is earning enough trust that we’d be happy to let that robot quietly take care of those jobs without thinking about it.

That isn’t an artificial intelligence problem.

It’s a human one.

The best robot may be the one you forget

I went to China expecting to see clever robots.

I did.

I saw the laboratories used to measure how effectively they clean, how much noise they create and how long their moving parts are likely to survive.

I participated in frank discussions about where existing systems still struggle, from changing garden environments to salt-covered windows and shallow swimming pools.

I watched how battery materials are coated, wound, filled, charged, aged and inspected before becoming the power source hidden inside a finished product.

What I did not expect was how much of the work would be devoted to making robotics feel unremarkable.

The public image of robotics is built around moments of visible intelligence: the machine identifies an object, makes a decision or completes a task that previously required a person.

Those moments may involve vacuuming a floor today. Tomorrow, they may involve preparing dinner or interacting with us in entirely new ways.

But those moments are only valuable when supported by thousands of less visible engineering decisions.

Will the seal survive?

Will the motor remain within its temperature limit?

Will the robot recognise the edge?

Will the battery behave consistently?

Can the product be repaired?

Will the company continue supporting it?

Can the owner trust what happens to the information it collects?

These questions are not as exciting as asking whether a robot has AI.

They are far more important.

The future of home robotics will not be determined solely by which company produces the most impressive demonstration.

It will be determined by which companies can make those machines dependable, understandable and easy to live with.

The best home robot is not necessarily the one that constantly reminds you how intelligent it is.

It is the one that quietly performs its task, returns to its place and becomes such a dependable part of the household that you eventually stop thinking about it at all.

The hardest part of building a home robot isn’t artificial intelligence.

It’s earning our trust.

Author’s note

Digital Reviews Network would like to thank Ecovacs Robotics for inviting us to Suzhou and Shanghai, and for the openness shown throughout the visit. The willingness of the engineering, product and leadership teams to engage in frank discussion made this editorial possible.

My thanks also go to Bianca Bai for extending the invitation, and to Laura, Lisa and the wider PR team for the enormous amount of planning and coordination behind what was my first international media visit. It’s easy to appreciate the presentations, factory tours and product demonstrations; it’s much harder to see the work required to make a week like this run so smoothly.

Finally, thank you to my fellow media attendees, Maggie, our interpreter, and our driver for making the experience such a memorable one.

A small part of the welcome awaiting the Australian and New Zealand media group in China.