Robots Atlas>ROBOTS ATLAS
Robotics & Hardware

LG Targets a Home Humanoid Around 2030, Appliances First

Lady Robot17 November 2026 · 6 min read
LG Targets a Home Humanoid Around 2030, Appliances First

LG Electronics has named the period around 2030 as its entry point into home humanoids, and said it will design household appliances so that robots can use them more easily. The company set out the plan on 28 October while discussing third-quarter results, as reported by Chosun Biz.

Key takeaways

  • Market order: research, public sector and industry first, households only around 2030
  • The central premise is a price at which a household would actually consider buying a robot
  • The robotics data factory has begun partial operations, with full operation in Seoul planned for the fourth quarter
  • An actuator pilot line was built in Changwon in the first half of the year, with production preparations under way in the second
  • Training-set target: roughly 500,000 hours of real and simulated data

Picture this situation

A humanoid is to take laundry out of a washing machine and move it to a dryer. For a person that is one action. For a robot it is a series of problems: find the door handle, pull it at the right angle, extract damp, tangled fabric from a deep drum, and avoid closing the door on its own wrist.

The washing machine was designed around a human hand and a human height. Nobody planned for a machine to open it.

But here is the problem: one can spend years improving a robot's manipulation, or redesign the washing machine once.

LG is pointing to the second route, and that is the most concrete part of this announcement. The laundry description above is an editorial example — the company did not say which appliances the change covers.

What did LG actually announce?

Lee Jae-mo, the LG executive responsible for robotics strategy, said the company expects the humanoid market to develop more fully around 2030. Its initial targets are academic, public-sector and manufacturing customers, with whom it will build component capabilities and deployment experience.

According to Chosun's account, Lee stressed reaching prices at which households would genuinely consider buying a robot. He described preparing robots designed for appliances and appliances designed for robots, drawing on the company's existing home-appliance expertise.

The implication is that LG can simplify household automation not only by improving the robot but by changing the device the robot has to deal with. The call set out neither a consumer price nor a firm launch date for a home humanoid.

Why is redesigning the washing machine easier than redesigning the robot?

Most companies' current approach is to raise manipulation capability until the robot copes with an environment designed for people. That route is expensive, because every new appliance in a home is a separate edge case.

LG holds an asset in this contest that no Silicon Valley robotics company has: it builds the appliances. Adding a handle a gripper can easily wrap around, or an interface through which a robot learns a cycle has finished, is a design change rather than a research problem.

The simpler option — waiting for manipulation models to grow into existing washing machines — falls short for two reasons. The first is time: nobody knows when that will happen. The second is reliability, because a home has no operator to correct the robot after a failed attempt.

There is also an awkward side to the plan. Co-design only works when the customer owns appliances of the same brand, bought recently enough. A home with a ten-year-old machine from another maker is back at square one.

Stage 1
Research hardware and actuators
Stage 2
Hardware-and-software packages per industry
Stage 3
Data-based services
Is the price acceptable to a household?
YES
Entry into the home marketAllow
NO
Keep driving cost downDeny
Home humanoidAllow

This is a simplified diagram built on the staged business model the company described, not an officially published roadmap with dates.

What already stands behind the plan?

LG's official earnings announcement confirms the robotics data factory has begun partial operations. The company also says preparations for actuator mass-production infrastructure are proceeding, alongside efforts to win orders from prospective global customers.

Financial Post's account of the same call adds that the actuator pilot line was built in Changwon during the first half of the year, with production preparations including equipment automation under way in the second. Full operation of the Seoul data factory is planned for the fourth quarter.

Popcorn News reports that LG is building a system combining real and simulated robot-training data, aiming to accumulate roughly 500,000 hours. Its collaboration with NVIDIA covers simulation, World Models and robot-development platforms. That figure describes an ambition, not confirmation of an already completed dataset.

The same report describes a staged business model: research hardware and actuators first, then industry-specific hardware-and-software packages, and finally data-based services. Paid testing services for outside customers are a longer-term possibility, while improving LG's own robots remains the immediate priority.

What do these names mean?

Three terms, without which the plan reads as a list of slogans:

  • Data factory — infrastructure for collecting and processing robot training data at scale, both real and simulated. The equivalent of a production hall, except what it manufactures is datasets.
  • Actuator — a motor assembled with a gearbox and sensors, meaning a single robot joint. It is the most costly repeated part of a humanoid, so in-house production decides the whole machine's cost.
  • Appliances designed for robots — a household device whose shape, handles or interface account for a machine operating it. The equivalent of accessibility in interfaces, only for a gripper instead of a screen reader.

The second and third bullets are linked here: in-house actuators lower the robot's cost, while robot-ready appliances lower the demands on its software. Both levers lead to the same end price.

Is it working yet?

No, this is a strategy announcement rather than a product launch. The company gave neither a consumer price nor a binding date for introducing a home humanoid, and framed 2030 as the expected point of fuller market development rather than a deadline for its own product.

A separate, nearer milestone is the early-2027 unveiling of a next-generation bipedal humanoid built on NVIDIA Isaac GR00T and Jetson Thor, which we covered in August. Showing a platform does not establish when households will be able to buy it.

What already exists is a partly running data factory and an actuator pilot line. The rest of the plan is a sequence the company intends to execute, not a present state.

Why it matters

Most humanoid players are trying to catch up with a world designed for people. An appliance maker can come from the other side and change that world, which is cheaper than perfecting manipulation. That is both an advantage hard for pure robotics firms to copy and a limit on the idea itself, because it works only in homes furnished by a single brand.

What's next

  • Full operation of the Seoul data factory is planned for the fourth quarter of this year
  • The next-generation bipedal humanoid on NVIDIA Isaac GR00T and Jetson Thor is to be unveiled in early 2027
  • The roughly 500,000-hour training-data target remains an ambition — only a description of the collected set will confirm it

Sources

Share this article