What a 10 Robot Hospital Trial Proves and Where It Stops

What did the China hospital trial find?

A six month study at a single tertiary hospital in Beijing compared 10 autonomous mobile robots with 19 manual workers on drug and specimen delivery. According to the peer reviewed paper in Scientific Reports, robot delivery reduced delivery time by 32% to 36% (p < 0.001). Over the six months, the robot fleet completed 7.3 times more delivery trips than the manual group.

The speed and accuracy results are strong. They still come from one hospital with its own elevators, corridors and staffing, so treat them as evidence that the approach can work in a busy setting, not as a forecast for your building.

How was the study designed?

The design was a six month, single center, parallel controlled study. Ten robots ran drug and specimen routes alongside a manual group of 19 workers, so both arms were measured over the same period.

Routing relied on Dijkstra's algorithm for global path planning, A* combined with a Dynamic Window Approach for local navigation, and an M/M/c queuing model to coordinate several robots at once. Queue behavior at elevators and doors shapes how much a fleet can actually move, so those design choices matter when you read the outcomes.

Why do the trip numbers differ between sources?

The brief's secondary source is a market research summary published shortly before our check. It describes 7.3 more delivery trips per week. The peer reviewed paper says 7.3 times more trips over six months. Those are different claims, and quoting the weekly version would overstate the trial by a wide margin.

The summary does not cite the paper. Its market figure of USD 9.75 billion by 2034 at a 15.7% CAGR is the summary's own estimate, drawn from 42 industry interviews, so it should not be treated as a verified forecast. Its claim of 615 medications and 415.8 staff hours saved in a 2025 pilot had no primary source in our check, so we leave it out.

How did accuracy and ramp up look over time?

Robot deliveries reached 100% verification accuracy and item integrity, with a lower omission rate (p = 0.011). Nurse and patient satisfaction were significantly higher in the robot group (p < 0.05).

Adoption was not instant. In the first month, average delivery time ran about 14% higher than in later months, and the authors report stabilization within 4 to 6 weeks. Their monthly table shows system uptime rising from 94.2% in June 2025 to 99.0% in November 2025. Plan for a rough first month, not a clean start.

What did the cost model claim, and what does it leave out?

The authors estimate that over a 10 year lifecycle the system saved 6.928 million RMB, and they describe their sensitivity analysis as confirming the economics. That figure is the authors' model output, not invoice data from a vendor or a hospital ledger.

Labor cost is the input most likely to differ in your market. A 2018 study by Yen et al., as cited in an arXiv routing paper, places food tray delivery and equipment retrieval at about 40% of nurses' time. That is a secondhand figure we have not checked against the original, so treat it as a question to answer with your own floor data.

Why do elevators decide the fit?

A separate prospective feasibility study at a tertiary hospital tracked 122 non urgent missions from June 18 to 29, 2025. Elevator operating rate was strongly associated with delivery failures, and higher elevator use lengthened delivery time. Most failures came from passengers or cargo physically blocking the robot.

The authors place an operational cutoff near 59.01% elevator operating rate. Below it, delivery success approximated 95.5%. Above it, failure was likely to rise sharply. That threshold tells a facilities team when to schedule runs away from peak lift traffic, which is more useful than any headline percentage. The study covered one hospital and one robot type, so the cutoff is a starting point for your own measurement. Singapore's Changi General Hospital offers another reference point, with a routing case study describing AMRs carrying drugs from a pharmacy to 32 wards twice daily.

What should US hospitals verify before applying the results?

Treat the trial as a template for questions, not as a set of targets. Before any US hospital cites these numbers in a plan, check the following against its own operation.

• Baseline: time stamped runner logs for drug and specimen routes, so the comparison starts from your own numbers rather than a generic manual average.
• Elevator data: operating rate by hour, mapped against the 59.01% cutoff the feasibility study reported.
• Route scope: which corridors, doors and wards are included, and which are excluded from the trip count.
• Uptime and fault response: documented uptime and the time needed to resume service after a stop.
• Specimen and medication handling: chain of custody and security requirements confirmed with your compliance team, since our research did not locate US regulatory guidance for this use.
• Staff acceptance: nurse and runner feedback collected before and after the pilot, not assumed.

Where does an integrator that leases fit in?

Service Robot Co. is a full service commercial robot integrator for US businesses. We are OEM neutral, so we select robots across manufacturers that fit a given site, then handle deployment, integration, training and service through a nationwide US engineer network. One partner covers the whole lifecycle.

Our offer is a lease. Robots are leased on a 3 year term with one fixed payment quoted after a site assessment. During that assessment, the questions above about routes, elevator traffic and baseline timing can be answered with real floor data before any payment is set.