Home Humanoid Repair: Service Network Checklist

China just gave the market a useful warning signal. JD.com launched a "robot ambulance" service for humanoid robots, quadruped robots, AI companion robots, and related machines. Jiemian reported that the service covers basic repair, fault diagnosis, battery replacement or recharging, testing and certification, cosmetic maintenance, and equipment recycling, starting in Beijing with a plan to expand to more than 50 core Chinese cities over three years. TechNode summarized the same launch in English.

That sounds like an odd news item until you picture the first wave of expensive robots in real homes. A 30 kg humanoid is not a phone. A 46 kg mobile manipulator is not a smart speaker. Even a 15 kg robot dog has batteries, joints, feet, sensors, radios, cameras, and software that can fail in ways a normal buyer cannot fix with a screwdriver.

So the practical question is not only "which robot should I buy?" It is: who will repair it where I live?

Who repairs a home humanoid robot when it breaks?

For now, the answer is usually one of four models:

1. Manufacturer direct support: the company repairs the robot, ships parts, handles diagnostics, and controls the warranty process.
2. Dealer or distributor support: a local reseller handles first-line service, especially for imported robots.
3. Platform service networks: a retailer, logistics company, or repair provider becomes the local field-service layer, like JD.com is attempting in China.
4. Owner/developer maintenance: common for research robots, where the buyer is expected to handle updates, calibration, accessories, and some repairs.

A home buyer should prefer the first three. The fourth can be fine for labs, universities, robotics teams, and developers, but it is a bad default for a family buying a robot to help with chores, companionship, or accessibility.

The more physical the robot is, the more the support model matters. A chatbot can be patched remotely. A humanoid can trip, bend a bracket, lose calibration, wear a gripper pad, damage a cable, swell a battery pack, or throw an error that prevents safe operation. A repair network is not a nice extra; it is part of the product.

The service problem gets bigger with humanoids

Look at the hardware ui44 tracks today. 1X NEO is a home-focused humanoid listed at $20,000 for early adopters, with a 167 cm body, 30 kg weight, around 4 hours of battery life, RGB cameras, depth sensors, tactile skin, microphones, Wi-Fi, and Bluetooth. That is exactly the kind of robot where a buyer should ask: if the battery degrades, a camera fails, or the robot needs safety recalibration, does a technician come to the home or does the whole unit ship back?

Unitree G1 makes the issue sharper because it is available now and much cheaper by humanoid standards. ui44 lists G1 at $13,500, with a 132 cm, 35 kg body, around 2 hours of battery life, 23 degrees of freedom, depth camera, 3D LiDAR, microphone array, Wi-Fi 6, Bluetooth 5.2, Unitree SDK support, ROS 2 compatibility, OTA upgrades, and optional dexterous hands on some variants.

That is a lot of machine for the money. It is also a lot of potential failure points. A low purchase price does not automatically mean low ownership risk. If you buy a research-style humanoid without a local support plan, you may be the integration department, the safety monitor, and the first-line repair desk.

This is why JD.com's "robot ambulance" signal matters beyond China. Chinese reports from IT之家 and Tencent said JD service had already supported more than 300 robots around testing and daily training for the Beijing humanoid half-marathon, fixing damaged or faulty machines so they could return to the training field. Tencent also reported a "120" after-sales standard: one-minute response, two-hour on-site service, and same-day repair, plus cooperation with robot companies including EngineAI, Noetix, Deep Robotics, and AGIBOT. Those are reported service goals, not a guarantee for every buyer or every geography, but they show where the market is going.

The first serious robot retailers will not only sell bodies. They will sell uptime.

Do not confuse availability with service readiness

A robot can be available to buy and still be hard to own. The checklist should separate five things:

The same robot may score well on some questions and poorly on others. An open-source research platform may have excellent documentation but limited consumer service. A polished companion robot may have a subscription plan but no third-party repair path. A low-cost humanoid may have a strong developer community but uneven support outside its home market.

Stretch 4 shows the assistive version of the problem

Hello Robot Stretch 4 is not a humanoid, but it is one of the clearest examples of why service should be treated as a spec. ui44 lists Stretch 4 at $29,950, available now for research, enterprise buyers, and in-home assistive pilot deployments. It has a 160 cm working height, 46 kg weight, 45 cm diameter footprint, self-charging, about 8 hours of light-load runtime, wide-FOV depth sensing, RGB cameras, hemispherical LiDARs, ROS 2 and Python SDK support, and a telescoping arm rated for 2.5 kg extended or 4 kg retracted.

Those specs are practical because Stretch is meant to reach things, not perform on a stage. But they also create real service questions. What happens if the arm needs calibration? Who replaces the lift mechanism? Can a caregiver safely restart the robot after an error? Does the home need a remote technician, a local robotics specialist, or a factory return?

Assistive robots raise the stakes. If a buyer with limited mobility depends on a robot to retrieve a phone, open a drawer, or bring an object, downtime is not just annoying. It may remove independence. Service level, backup plan, and data handling should be discussed before the pilot begins.

Stretch also shows why repairability is not only about humanoids. The first useful home robots with arms may be wheeled, slow, and specialized. That makes them easier to stabilize than bipeds, but it does not make them maintenance-free.

Robot dogs expose the battery and parts question

The same service-network logic applies to robot dogs. JD and Chinese media framed the robot-ambulance launch broadly enough to include quadruped robots, not just humanoids, which is the important buyer signal. Unitree Go2 is one of the few robot dogs that a technically curious buyer can realistically purchase today, so it is a useful example from ui44's database without needing to assume a confirmed model-specific JD service listing.

ui44 lists Go2 from $1,600 for Go2 Air, $2,800 for Pro, and $4,500 for X, with EDU pricing by contact sales. It is around 15 kg, has 1-2 hours of standard runtime or 2-4 hours on the EDU long-endurance battery, uses 4D LiDAR, cameras, IMU, Wi-Fi 6, Bluetooth 5.2, optional 4G/LTE, OTA software updates, voice commands on some trims, and payload ratings up to 8-12 kg.

That price range makes robot dogs feel more approachable than humanoids. But a legged robot still has high-wear parts: foot pads, leg joints, shells, batteries, chargers, sensor covers, and sometimes cellular modules. If it falls on stairs, gets wet, damages a joint, or loses calibration, the repair path matters more than the launch video.

For buyers, the Go2 lesson is simple: do not stop at the checkout page. Ask whether batteries are replaceable, whether spare feet and shells are stocked, whether a local service partner exists, and whether a fault can be diagnosed without uploading full logs, maps, images, or location history.

Companion robots already sell service plans

Companion robots make the support question less dramatic but more emotional. LOVOT is a good example because GROOVE X treats care as part of the product. ui44 lists LOVOT 3.0 at ¥577,500 in Japan with a required care plan starting from ¥9,900/month. The robot is only 43 cm tall and 4.6 kg, but it has more than 50 sensors, person recognition, room mapping, touch sensors, OLED eyes, NFC, infrared communication, Wi-Fi, Bluetooth, a charging nest, and 30-45 minutes of active runtime between nest visits.

A broken LOVOT is not a 35 kg safety hazard. But it may be attached to a child, an older adult, a store, or a care setting. Repair is continuity. If the company can repair the body but cannot preserve settings, behavior, clothing profiles, or relationship state, the buyer should know that upfront. If parts become scarce, the support plan should say what happens next.

ElliQ 3 is different again. It is stationary, U.S.-only, and subscription-based. ui44 lists a $249 lease initiation fee plus membership plans such as $49/month monthly or annual/24-month options, with a camera, four-microphone array, touchscreen, wellness programs, medication reminders, video calling, memory recording, and caregiver dashboard. For ElliQ, support is less about shipping a heavy robot body and more about service availability, account continuity, caregiver access, data export, and what happens when a subscription ends.

This is the broader lesson: a service network is not only a van with tools. It is also account recovery, privacy handling, spare parts, support geography, and an end-of-life path.

How the current robots compare

Here is the practical read from ui44's database.

No single service model fits all of these robots. A $1,600 robot dog, a $20,000 home humanoid, a $29,950 assistive manipulator, and a subscription-based senior companion have different failure modes. But every serious buyer should ask the same underlying question: is the repair path as real as the demo video?

Bottom line: service is now a home robot spec

JD.com's robot ambulance is not proof that every home robot is suddenly easy to repair. It is one early sign that the robotics market is growing out of the demo phase. When companies sell humanoids, quadrupeds, and AI companions to normal buyers, they inherit normal buyer expectations: quick response, local service, clear warranty terms, replacement parts, data protection, and a plan for old hardware.

For ui44 readers, the buying rule is simple. Do not evaluate a home robot only by height, payload, walking speed, AI model, or preorder price. Add a service row to your comparison sheet. If the company cannot explain who repairs the robot, where parts come from, how batteries are replaced, what data technicians can see, and what happens after warranty, the product is not ready for your home yet.

A home robot is not just a machine that moves. It is a machine that will sometimes stop moving. The best robot companies will be the ones that planned for that before the buyer had to ask.