When a team starts looking for an indoor delivery robot, the first question is almost always the same: how much can it carry? It is the right question, and it is also the one that hides the ones that decide whether the project works. A delivery robot payload is a specification on a page. What actually determines whether the machine can do the job is whether the building accepts it — the lift, the doorway, the threshold, the floor and the person who has to load it.
This guide is about the second half of the problem. It covers what a payload figure does and does not tell you, which parts of the building have to be checked against it, and the survey worth running before a single machine is ordered. Our indoor delivery robots are documented the same way: what the platform is for, and which site questions we need answered before recommending it.
Payload is a volume problem before it is a weight problem
A rated payload is expressed in kilograms, but almost nothing that moves through a hotel, hospital or office is limited by mass. Meal trays, linen bags, sealed pharmacy orders, documents and specimen containers are all bulk-limited: they run out of space in the compartment long before they approach the weight limit.
That has a practical consequence. Two machines with the same payload rating can have very different useful capacity, because the number of tray positions, the internal height and the way the compartment opens decide how many round trips a shift actually needs. When we quote a delivery platform, the question that matters is not “how many kilograms” but “how many of the thing you actually carry, per trip.”
The same logic applies in reverse on the industrial side of the same product family. An AMR moving totes and racks is specified by load and by how that load is picked up and set down, not by a single weight ceiling.
What a delivery robot payload actually has to fit
Before comparing models, write down what has to travel. Four things turn a vague requirement into a specification:
- The unit of delivery. Is the machine carrying a sealed tray, a bag, a stack of documents or a loose item? The unit decides the compartment shape you need, not the other way round.
- The worst-case item, not the average one. Most buildings have one awkward load — a catering box that will not stack, a linen bag that will not compress. If the machine cannot take that one, the route still needs a person.
- Whether the load is sealed. Compartments that lock are specified for pharmacy, specimen and valuable deliveries; an open tray is fine for meal rounds. This is a compliance question as much as a design one.
- How many trips a shift needs. A machine that holds two trays is not wrong; it is wrong if the round needs ten trips and the shift has room for four.
A delivery robot payload figure is therefore a starting constraint rather than an answer. The model pages in this range — Shan Dian, De Li Ge, Panda and Amy — describe the service each platform is intended for, and we confirm the details against your actual load rather than asking you to read them off a table.
The lift decides more projects than the robot does
In a multi-floor building, the lift is the constraint that most often kills a delivery robot project after the decision has already been made. Three separate things have to be true, and all three are about the lift rather than about the robot:
- Weight. A lift has a rated load, and it is the load on the car, not the machine on its own, that matters — plus whatever is standing in there with it.
- Free floor area inside the car. A machine can pass a door and still not be able to turn inside a lift.
- Who gets priority. An autonomous machine that cannot call a lift independently needs either a person to press the button or a building system that will talk to it. That is an integration decision, not a robot feature.
If the answer to the third point is “nobody has agreed yet,” that is the first conversation to have. A hospital route that ends at the lift lobby is not a delivered route; it is a route with a gap in it. Where the site genuinely cannot give the machine lift access, the honest answer is usually to keep the deployment on one floor or in one building, and to say so before the order rather than after.
Doors, thresholds and the last metre
Everything between the corridor and the destination is a potential stop. The list is short, and every item on it is measurable in an afternoon:
- Clear door width of the narrowest door the route passes — fire doors, ward doors and service doors are usually different widths.
- Thresholds and expansion joints, which behave differently from a smooth floor and can be the reason a machine refuses a route it was mapped for.
- Ramps and floor changes, including any incline between a kitchen and a dining area or between a loading dock and a corridor.
- Lift door timing — how long the doors stay open, and whether that is enough for a machine to enter, position and clear.
- Final handover point: whether the machine can stop beside a desk, a ward station or a table without blocking it.
None of these require special equipment to check. A tape measure and a walk of the actual route will find almost all of them, and finding them before delivery is what keeps a deployment to one visit instead of three.
Loading and unloading: who is doing it?
The most overlooked part of a delivery robot project is the human step at each end. A machine does not load itself, and the whole labour saving depends on whether the loading is quick enough to be worth doing.
In practice this is where the design gets decided. A hotel meal round works because trays arrive on a trolley and slide across in seconds. A hospital pharmacy route works because sealed orders are prepared in the pharmacy and collected at the unit. A route where somebody has to pack the machine item by item will be slower than the walk it replaced, and will quietly stop being used.
So the question to ask at each end of the route is the same: who loads it, who unloads it, and how long does that take compared with doing the trip by hand. If the answer is not obviously faster, the machine is being bought to solve a problem it cannot solve. The delivery scenarios already documented cover this at the route level for restaurants and food service and for hospital internal deliveries.
The survey to run before you order
Nine questions, in this order. The answers decide the platform; the platform does not decide them.
- What are the three most common things that will travel, with their real dimensions?
- What is the single most awkward item, and can the route be planned around it?
- How many trips does the busiest shift actually need?
- What is the narrowest clear door width on the route?
- What is the lift rated load and the free area inside the car, and has lift priority been agreed?
- Are there thresholds, ramps or floor changes on the route?
- Does any load need to travel in a locked compartment?
- Who loads and who unloads, and where does the machine stop at each end?
- Who owns the route once the machine is running?
If you can answer those, the machine choice is a short conversation. If you cannot, no payload figure will help — and a comparison of AMR and AGV approaches will not either, because the difference between them is a routing and layout decision rather than a carrying-capacity one.
How to start
Send us the three most common loads with their dimensions, the busiest route with its narrowest door and lift details, and the hours the machine is allowed to work. From those we will suggest a platform, the compartment arrangement and the loading points, and tell you plainly if the route needs a change before it will work. Send the details here, or compare the classes in the machine overview first if you would rather shortlist on your own.