Care Robots at Home: Hospital Deployment Lessons

That is why hospital and care-facility deployments matter. They do not prove that a robot is ready to replace a caregiver at home. They reveal the minimum standard: reliability, clear workflows, safe navigation, support contracts, and a plan for what happens when the robot gets confused.

The most useful signal in 2026 is not a humanoid promising to do everything. It is a care robot doing a narrow job often enough that staff stop treating it like a novelty. F&P Robotics' Lio and Diligent Robotics Moxi are good reality checks for anyone evaluating home care robots.

What hospital deployments actually prove

In its November 2025 deployment report, F&P Robotics says Lio travelled more than 1,000 km in 217 days at University Hospital Basel with over 99% reliability and an ROI of 0.78. The short company report does not explain how it calculates reliability or return on investment, so those figures are a starting point for questions, not a home-care performance guarantee. Its HRI 2026 white-paper announcement describes multi-year use in a rehabilitation clinic and a more recent university hospital laboratory deployment, focusing on trust, task changes, safety, regulation, and organisational workflow rather than raw robot hardware.

Those details are more meaningful than a single flashy demo because they describe operation over time. A care robot has to find the right room, avoid people in hallways, recharge, recover from blocked paths, fit into staff routines, and be maintained without becoming another burden.

Lio's official product page frames it as a mobile service robot with a collaborative arm, voice interaction, autonomous navigation, an integrated display, and configurable workflows for healthcare, elderly care, laboratories, and education. The 2020 Lio research paper adds useful engineering context for that version: soft body covering, collision detection, limited speed and forces, visual/audio/laser/ultrasound/mechanical sensors, autonomous recharging, up to 8 hours of battery life, and local onboard computing for privacy-sensitive care environments. The paper also says Lio complies with ISO 13482 safety requirements for personal-care robots. That is an engineering safety claim, not certification that it improves clinical outcomes.

That still does not make Lio a plug-and-play home caregiver. It means Lio has been built around the boring things homes will also need: repeatable tasks, safety constraints, maps, maintenance, and local trust.

Moxi tells a similar story from the hospital logistics side. In its acquisition announcement, Diligent reported nearly 100 robots, more than 1.25 million autonomous deliveries, and deployments across 25+ U.S. hospital facilities. Its job is deliberately narrow: supplies, medications, lab samples, PPE, central-supply runs, and other non-patient-facing tasks. Diligent Robotics' product materials emphasize that Moxi can be implemented without major infrastructure rebuilds, over existing Wi-Fi, with workflows customized by an implementation team.

That is the lesson for homes: the robot is not the product by itself. The workflow is the product.

The home lesson: care is a workflow, not a feature list

A private home is less regulated than a hospital, but it is not necessarily easier. Hospitals have trained staff, clear corridors, known rooms, service contracts, and people who are paid to adapt workflows. Homes have stairs, slippers, pets, emotional privacy, family preferences, kitchen clutter, and a person who may not want to troubleshoot a robot after dinner.

For buyers, the care-robot question should be: which part of the care workflow is this robot actually taking responsibility for?

A home robot does not need to do all five. In fact, the safest early products will usually do one or two jobs well.

Compare care robots by job, not by shape

The ui44 database shows why "elder-care robot" is too broad. Some products are stationary companions. Some are mobile screens. Some have arms. Some are facility robots that may influence home design later. These should not be ranked as if they solve the same problem. Prices and sales routes below were checked on October 14, 2026; manufacturer specifications are not clinical trial results.

This comparison leads to a less exciting but more useful answer: near-term care robots are not replacing caregivers. They are splitting care into smaller jobs.

Social care is the most home-ready category

ElliQ 3 is one of the clearest home examples because it does not pretend to be a nurse. Its official specifications describe a stationary companion robot for older adults, about 14.4 inches tall, 5.3 lb, mains-powered, with a camera, far-field microphones, an 8-inch touchscreen, Wi-Fi, Bluetooth, ElliQ Voice AI, and a caregiver app. Its functions are reminders, conversation, health and pain tracking, video calling, virtual activities, photo sharing, games, music, and wellness programs. Health tracking means logging readings and pain the user reports, with optional caregiver notifications; it is not clinical sensing or an emergency response service.

That is a real care workflow, but a limited one. ElliQ can help structure a day, reduce friction in communication, and support routine prompts. It cannot pick up a dropped phone, bring water from the kitchen, check whether someone has fallen behind a closed door, or physically assist with bathing or transfers.

LOVOT and PARO sit on the emotional-support side of the market. In Japan, LOVOT 3.0 in the basic ちゃ color costs ¥577,500 including tax, administration and shipping, plus required service from ¥9,900/month, as checked October 14, 2026. The hardware price applies through October 25; ¥599,500 starts October 26. Its 3.0 specifications include touch sensors, room mapping, person recognition, temperature/humidity sensing, and a 30–45 minute active cycle before returning to its nest. The thermal camera listed for LOVOT 2.0 should not be assumed for 3.0.

PARO is a therapeutic seal robot used in hospitals and elder-care facilities. Its manufacturer describes tactile, light, audio, temperature and posture sensors, with microphones for sound. The FDA device listing identifies PARO as a Class II biofeedback device. That listing is not evidence that PARO can provide independent home care.

These robots are useful precisely because they do not try to manipulate the world. Their risk is different: emotional dependency, consent, privacy, and whether the person actually wants a robot companion.

Physical help is still the hard part

The moment a care robot has to move objects, the bar rises sharply. Carrying a blanket, opening a drawer, lifting a cup, or finding glasses on a cluttered table requires perception, reach, grasping, force control, safety limits, and recovery when the first attempt fails.

Stretch 3 shows the most honest version of that problem. Hello Robot lists a $24,950 price. The robot weighs 24.5 kg, has a compact 33 × 34 cm footprint, reaches 141 cm tall, carries a 2 kg payload, runs 2-5 hours, and supports ROS 2, a Python SDK, web teleoperation, and embodied-AI research. It is built for real homes and assistive research, but the buyer should notice the framing: research platform, teleoperation, open-source autonomy, and controlled tasks. That is not the same as a retail robot caregiver.

Robody takes a different route. Its advertised €690/month all-inclusive home-care service remains a preorder/waitlist offer; coverage for a particular home needs confirmation. Devanthro lists 1.65 m height, 60 kg weight, 6-hour battery life, 5G/Wi-Fi, mm-wave radar, 4K fisheye cameras, stereo microphones, 1.5 kg per arm payload, self-docking, fall-alert workflows, and human-in-the-loop VR teleoperation for tasks requiring judgment or dexterity. That human backup is not a weakness. For care, it may be the feature that makes the product plausible.

Facility robots like aeo point in the same direction. Aeolus announced dual 7-DOF arms, an 8 lb / 3.6 kg single-arm lift capacity, autonomous elevator and door operation, object pickup, delivery workflows, security patrol, and UV disinfection; its current product page also lists mobile-app access and flexible leasing. It is not a consumer home robot, but it shows how manipulation becomes more credible when paired with bounded facility tasks and a service model.

What buyers should ask before trusting a home care robot

A care robot purchase should start with operational questions, not brand demos.

1. What exact job will it do every day? A reminder robot, telepresence robot, mobile manipulator, therapeutic pet, and hospital logistics robot solve different problems.
2. What happens when it fails? Look for human backup, remote support, caregiver escalation, clear error messages, and a way to pause or override the robot.
3. What does the robot do locally? Care environments involve sensitive data. Lio's older technical paper is notable because it emphasizes onboard compute for privacy-sensitive algorithms; home buyers should ask the same cloud/local question.
4. Who maintains it? Hospitals buy support. Families often buy devices. A home care robot without a service plan may turn into another object someone has to care for.
5. Can the person say no? Consent matters. A robot that prompts, records, follows, or watches someone should be easy to mute, park, or restrict.
6. Does it fit the home? Measure doors, thresholds, rugs, furniture gaps, Wi-Fi coverage, charging location, lighting, and caregiver access.
7. Is the evidence about care outcomes or task completion? A robot completing deliveries is good evidence for logistics. It is not evidence that the same robot improves dementia care, prevents falls, or replaces a nurse.

The best use of ui44 compare is therefore not to find "the smartest care robot." It is to compare the job boundaries: stationary companion, telepresence, mobile manipulator, facility robot, or therapeutic companion.

So, are care robots ready for homes?

Some are ready for narrow home roles. ElliQ-style companionship, reminders, wellness prompts, video connection, and structured daily engagement are the most realistic. Therapeutic companions such as PARO can be useful in care contexts when the goal is calming interaction rather than independence. Telepresence and monitoring robots can help families check in, if the person being monitored wants that presence.

Physical care is earlier. Stretch 3, Robody, Lio, Moxi, and aeo all show pieces of the puzzle: manipulation, navigation, teleoperation, workflow integration, safety limits, and support models. But the evidence is strongest in labs, hospitals, care facilities, and supervised deployments, not unsupervised private homes.

That is not a pessimistic answer. It is the path to a better one. Care robots will become useful at home when they stop being sold as general caregivers and start being evaluated like service systems: one job, one workflow, one support model, one failure plan, repeated every day.

For now, hospital deployments reveal the right buyer posture: be optimistic about narrow assistance, skeptical about broad caregiving claims, and very interested in any robot company that can show long-term reliability outside a demo room.