Start with a load that makes sense
Hospital logistics covers many different journeys. Some involve predictable replenishment, such as gloves, dressings or clean linen. Others carry medicines, diagnostic samples, meals or equipment. These are not interchangeable tasks. A suitable container, a clear destination and a documented receiving process matter more than whether a robot can physically carry the weight. Start by describing the service that people need, then assess whether an autonomous delivery can support it.
Routine, repeatable movements can be useful candidates for an initial review because their volumes and destinations are easier to understand. That does not make them automatically safe or economical. A ward with very little storage may need small, well-timed deliveries rather than a large drop. A distribution room may have a narrow doorway or limited loading space. Staff who prepare and receive supplies should help define these practical limits before a route is selected.
Preparation is a human checkpoint
Before dispatch, someone must confirm the correct item, quantity, destination and handling requirements. Barcode scanning may support those checks, but scanning is not proof that the box contains what the label describes. An organisation needs a process for substitutions, damaged packaging and missing items. Where traceability is required, it should be designed around the goods and local rules rather than assumed to come with the robot.
GS1 standards provide established approaches to identifying products and locations. NHS Supply Chain is a useful reference point for understanding the wider supply environment in England. Neither organisation’s name on a checklist makes a particular delivery workflow compliant. The hospital still needs to decide which identifiers it uses, who maintains the data and what happens when a label cannot be read.
Match the container to the contents
Some robots carry enclosed compartments; others tow or lift compatible carts. Payload ratings, stability limits and docking requirements vary by product. The approved arrangement should consider the loaded centre of gravity, movement on ramps and what happens during a sudden protective stop. A container that fits on a platform is not necessarily approved for that platform. Loose items, trailing straps and poorly closed drawers create avoidable risks.
Clean and used materials need appropriate segregation. A vehicle used for more than one service requires an agreed cleaning process, including responsibilities, compatible products and access to frequently touched surfaces. The World Health Organization’s infection prevention and control resources provide broader context, while local infection prevention teams determine what is suitable for their environment. Transport automation does not remove contamination risks or replace cleaning verification.
Some loads need additional controls
Medicines may require restricted access, temperature management and records of custody. Pharmacy leadership must approve the actual process, including delays and failed receipt. The General Pharmaceutical Council’s standards are relevant context for registered pharmacy services in Great Britain. A locked robot compartment alone does not establish that a medicine has been stored or delivered correctly, and controlled drugs require particular attention to applicable law and local procedures.
Specimens introduce different questions: suitable packaging, orientation, time sensitivity, temperature and response to leakage. ISO 15189 concerns quality and competence in medical laboratories; it is not a blanket approval for robot transport. Where dangerous goods rules apply, requirements such as UN 3373 and Packing Instruction P650 may be relevant. Those requirements depend on classification and transport circumstances, so laboratory and dangerous goods specialists should review the proposed journey rather than relying on a generic checklist.
Receipt finishes the service
An arrival notification should tell an authorised person what action is needed without exposing unnecessary clinical information. The receiving colleague checks the load, notes exceptions and confirms acceptance through the agreed method. If no one is available, the system needs an explicit waiting, return or escalation policy. Leaving a cart in a corridor is not a neutral outcome: it can obstruct movement, compromise security and shift unplanned work onto clinical staff.
Finally, supplies must reach their intended storage location. A successful transport event is not necessarily successful replenishment. Measure incomplete receipts, time spent finding a delivery and stock that remains unavailable after arrival. These observations reveal whether the service genuinely supports care. The most useful improvement may be a clearer receiving shelf or a better schedule, rather than a faster robot.