Ten Autonomous Mobile Robots Cut Hospital Delivery Time 32% to 36%

The Delivery Bottleneck on Every Ward

Drug and specimen runs in a tertiary hospital form a constant, time sensitive flow. Every medication box and sample container has to reach the right ward intact, and every trip ties up a person who could be doing other work. Manual delivery carried that load, which left turnaround times dependent on how the day unfolded.

The trial was built to make the comparison fair. Manual delivery ran as a parallel control against the robot group for six months, so both methods were measured inside the same hospital. The question for the hospital was whether a robot fleet could match the speed, accuracy and staff effort of its existing team.

Coverage made the problem harder. The fleet had to serve 22 open wards, so the test was not whether a robot could travel a single corridor. It was whether a fleet could hold service levels across the whole building at peak hours.

Fleet Sizing and Rollout

Fleet size was set from a peak hour demand analysis and service level targets rather than a guess. The resulting fleet of 10 autonomous mobile robots served the 22 open wards. Each unit measures 785 mm by 540 mm by 1430 mm, weighs 145 kg, carries a 200 kg payload and can take several medication boxes or sample containers on each trip.

Navigation combined global and local planning. Dijkstra's algorithm handled global path planning, A* with the Dynamic Window Approach managed local navigation, and an M/M/c queuing model coordinated the robots as a group. Each robot runs on a 24 V 30Ah lithium ion battery and operates for about 7 hours per charge.

The study states that the system can reach maturity within 4 to 6 weeks after deployment, which sets realistic expectations for the period between go live and steady performance. Staff roles were reorganized as well, with the 10 robots taking over work previously handled by 19 full time logistics personnel.

• Size the fleet from peak hour demand and service level targets
• Plan global routes and local obstacle avoidance for each unit
• Coordinate the fleet as a group so trips are shared across wards
• Run a parallel comparison against manual delivery for six months
• Expect steady performance within 4 to 6 weeks after deployment

Measured Results Across Six Months

Delivery time fell 32% to 36% against manual delivery, a difference the study reports at p < 0.001. The robot group also reached 100% verification accuracy and item integrity, and its omission rate was significantly lower (p = 0.011).

Monthly data show the fleet settling in. Trips rose from 10,500 in June 2025 to 14,204 in October 2025. Average delivery time moved from 23.9 minutes in June to 21.0 minutes from July onward, and the error rate fell from 0.012% to 0.003%. Uptime climbed from 94.2% to 98.7% over the same five months.

Staff and patient experience moved in the same direction. Nurse and patient satisfaction were significantly higher in the robot group (p < 0.05). Over six months the fleet completed 7.3 times more delivery trips than 19 manual workers, and the study's own economic analysis puts ten year savings at 6.928 million RMB.

What This Means for Facilities Weighing Robots

Service Robot Co. is a full service, OEM neutral commercial robot integrator for US businesses. We select the right robots across manufacturers, then finance, deploy, integrate, train and service every unit through a nationwide US engineer network, so one partner covers the whole lifecycle. A facility weighing a delivery fleet can start with a site assessment before any hardware is chosen.

The questions this trial answers are operational: how many units match peak demand, how long a fleet takes to reach steady performance, and who keeps it running. Those are the same questions a site assessment works through. We lease robots on a 3 year lease with one fixed payment quoted after that assessment, so the decision rests on the operation rather than on the upfront cost of hardware.

The trial also marks its own limits. Its results come from one tertiary hospital over six months, and the findings are not presented as a forecast for other buildings. Any deployment should be sized against its own floors, elevators and demand profile.