Can a Humanoid Robot Load the Dishwasher?

Figure's January 2026 Helix 02 post is the strongest public dishwasher-loading claim so far: a Figure robot completes a roughly four-minute autonomous kitchen sequence, walks between the dishwasher and cabinets, unloads dishes, stacks items, reloads the dishwasher, and starts it using onboard sensors with no human intervention or resets. That is meaningful. It is also not the same thing as a consumer appliance.

The useful way to read the demo is as a benchmark. Dishwasher loading combines almost every capability that matters for home robotics: perception, hands, balance, two-arm coordination, force control, task memory, and recovery when a cup slips or a rack gets in the way. If a robot can do this repeatedly in an ordinary kitchen, it is much closer to being a useful home helper.

But most robots in the ui44 database are not there yet. Even the most interesting platforms are split between impressive autonomy demos, incomplete consumer support, missing public specs, and robots that are research tools rather than finished household products.

Can a humanoid robot load the dishwasher?

The honest answer is: in controlled demonstrations, yes; as a dependable home product, not yet.

Figure has shown the clearest public evidence. Its earlier "Helix Loads the Dishwasher" post explained why the task is harder than it looks: dishes start in cluttered stacks, objects must be isolated and reoriented, slippery glasses need fingertip-level precision, dishwasher racks allow only centimeter-scale error, and every load creates new collisions and edge cases. Figure says Helix learned this with the same general architecture used for towels and packages, with new data rather than special-case code.

Helix 02 pushes that from a manipulation clip into full-body autonomy. Figure says the system connects onboard vision, touch, and proprioception to whole-body control; in the dishwasher sequence it performs 61 ordered loco-manipulation actions over several minutes. The most interesting details are not the dishes. They are the whole-body tricks: walking while holding delicate objects, using both hands to transfer and stack items, closing a drawer with a hip, lifting a dishwasher door with a foot, and keeping task state over a long sequence.

That is exactly why buyers should be careful. A four-minute uncut demo is much better evidence than a jump cut. It still does not answer day-one questions: What kitchen layouts were tried? How often did it fail? What happens with wet plates, knives, pets, kids, dark glass, unusually shaped bowls, or a rack that is already half full? Can the owner intervene safely? Who is liable if it breaks something?

Why dishwasher loading is a serious home-robot benchmark

Dishwasher loading is a better benchmark than a robot waving, dancing, or carrying a box across a factory floor because it compresses the home-robot problem into one recognizable chore.

A useful dishwasher robot has to solve six things at once.

First, it needs perception that works in clutter. A white plate on a white counter, a transparent cup, a reflective spoon, or a bowl partly hidden behind another bowl can confuse systems that look fine on clean tabletop demos.

Second, it needs hands that can grip without crushing. The robot has to handle ceramic, glass, metal, plastic, and possibly wet surfaces. A gripper that can lift a package is not automatically safe around stemware.

Third, it needs bimanual coordination. Many dishes need to be reoriented before they fit. That means one hand may stabilize while the other rotates, or the robot may pass a cup from one hand to the other.

Fourth, it needs whole-body balance and reach. The rack is low. Cabinets may be high. The dishwasher door changes the robot's stance. If the robot has to walk while holding dishes, locomotion and manipulation cannot be treated as separate jobs.

Fifth, it needs sequencing. "Load the dishwasher" is not one command; it is a long chain of actions with dependencies. Open the rack. Pick the plate. Check orientation. Place it. Avoid the previous plate. Repeat. Close the drawer. Start the machine.

Sixth, it needs recovery. A demo where every grasp works is less interesting than a system that notices a misgrasp, changes strategy, and continues without a human reset.

That is why dishwasher loading is a useful filter for humanoid claims. It is not because everyone hates dishes more than laundry. It is because the chore exposes whether a robot can deal with real homes rather than scripted props.

What the ui44 database says about likely contenders

Figure is the demo leader, but it is not the only robot worth comparing. The ui44 database currently tracks several platforms with arms, hands, household positioning, or manipulation hardware that matter for this chore.

The pattern is clear: the robots closest to consumer purchase are usually not the ones with the strongest public dishwasher evidence. The robots with the strongest full-body manipulation story are not ordinary consumer products.

This is why 1X NEO is both exciting and a useful cautionary case. At $20,000, it is one of the clearest home-humanoid purchase signals in the database. It is 167 cm tall, only 30 kg, rated at about four hours of battery life, and designed around a soft body, tactile skin, RGB/depth sensors, and household chores.

But 1X also talks about Expert Mode: if a chore is not known, a 1X Expert can guide the robot while it learns. That may be a practical bridge for early home robots. It is also not the same as fully autonomous dishwashing. Before buying, you would want to know which chores are autonomous on day one, what needs remote human help, what video or sensor data leaves the home, and how failures are handled.

Sunday Memo is even more directly aimed at this exact problem. Sunday says Memo can clear tables, handle delicate glasses, throw out food scraps, load the dishwasher, fold laundry, and make coffee, with a beta planned for late 2026. That is the right target. The missing piece is public hardware specificity: no official retail price, height, weight, runtime, sensor array, payload, or safety certification details are available in the ui44 record yet.

Stretch 3 is the opposite: fewer but more concrete promises. It is a $24,950 mobile manipulator with a 2 kg payload, RGB-D cameras, LiDAR, ROS 2 support, Python SDK support, and 2-5 hours of battery life. It is valuable because it proves a grounded point: a non-humanoid mobile arm can already be a serious platform for home manipulation research and assistive care. But it is not a sealed consumer product that will quietly do the dishes after dinner.

What should buyers ask before trusting a dishwasher demo?

A dishwasher video should make you ask better questions, not fewer questions. Use this checklist when comparing Figure, 1X, Sunday, Unitree, AGIBOT, NEURA, Tesla, Apptronik, X Square, or any other home-robot claim.

Is the demo autonomous, teleoperated, or assisted?

Figure explicitly says the Helix 02 dishwasher sequence used onboard sensors with no human intervention. That is the right kind of claim to look for. If a seller says "AI-assisted," "remote support," "expert mode," or "operator in the loop," ask exactly when human help appears and whether you will know when it is active.

How many layouts and loads were tested?

One kitchen can hide a lot of brittleness. A good dishwasher benchmark should include different rack designs, counter heights, cabinet locations, lighting, wet dishes, transparent glass, clutter, awkward pots, and partially filled racks. The more a demo depends on a pre-arranged scene, the less it tells you about your home.

What are the hands actually rated to do?

Look past the word "dexterous." Ask about payload across the full arm range, fingertip force control, tactile sensing, grip materials, hand degrees of freedom, and what happens when the object slips. ui44 tracks payload details because they matter. AGIBOT X2, for example, has a 3 kg maximum in specific postures but a 1 kg full-range payload. That difference is huge in a kitchen.

Can it recover without a reset?

Dishwashers are recovery machines. Plates lean. Cups tip. Silverware catches. The most valuable robot behavior is not perfect first-try placement; it is noticing failure, changing grip, moving an obstacle, or deciding that an item is unsafe to handle.

What support model comes with the robot?

A home robot is not just hardware. Ask about warranty, spare parts, software updates, cloud dependency, remote-human involvement, emergency stop behavior, liability, data retention, and whether the robot can be serviced locally. Use /compare to separate robots with similar marketing language but very different size, price, battery, payload, and status signals.

The bottom line

Figure Helix 02 makes dishwasher loading feel real in a way older humanoid demos did not. The important advance is not that a robot touched dishes. It is that Figure presented a multi-minute, whole-body, autonomous sequence that combined walking, grasping, balance, two-arm coordination, task memory, and recovery. That is exactly the direction home robotics needs to go.

For buyers, the answer is still cautious. If you want a robot that can reliably load your dishwasher this year, the market is not there. If you want to track which companies are getting closest, watch for three signals: uncut autonomous home demos, published failure/recovery rates, and concrete product support for ordinary households.

Until those appear together, dishwasher loading is best treated as a benchmark, not a promise. It tells you which robots are becoming more capable. It does not yet mean the dishes are off your chore list.