Published specifications are accurate and measured under conditions chosen to produce good numbers. The useful skill is not doubting them but knowing how to translate each one into your conditions.
Battery and operating time
The most commonly misread figure across every robot category.
How it is measured. Typically in a low-power mode: screen dim or off, audio silent, minimal movement. For cleaning robots, on the lightest setting over open floor.
Real operation is different. Bright screen, audio active, continuous movement, and for cleaning robots, water and suction running.
What to ask. Operating hours with the screen bright, audio on and moving continuously. That figure is usable.
Charging time matters equally. Four hours of operation with six hours of charging means one session per day, which may not fit the requirement.
Ask about automatic docking. A robot that returns to charge itself is operationally different from one requiring someone to plug it in, and where all-day operation is needed the second becomes a real burden.
How to verify. Request a run from full charge until it must recharge, under conditions resembling actual use. The number from that test beats every published figure.
Payload and capacity
The figure most often causing sizing errors.
Rated payload includes the tool. Gripper, tool changer and cabling can consume a large share on smaller arms, leaving considerably less for the actual item.
Centre of gravity offset derates it further. Ratings assume the load sits close to the mounting flange. An item held well out loads the wrist far more, and the derating is substantial.
Inertia is rated separately. A light but bulky item can exceed inertia limits while well under the mass limit. The symptom is accelerated wear and position error at speed rather than an error message.
For carrying robots, capacity assumes ideal loading. Heavier loads mean slower travel and shorter runtime.
A working rule for early sizing. Assume roughly half the rated payload is usable for the item itself, then refine with the supplier's load calculation using real figures.
And size up early. Moving to the next size costs far less at quotation than discovering the problem during commissioning.
Throughput and speed
Figures that describe a test rather than your application.
Cycle times reference a standard test move. Usually a short defined path with a specified payload, in ideal conditions.
Your cycle differs in ways that all reduce it. Heavier payload, longer or more complex path, settling time before the gripper acts, signal handshaking, and waiting for other equipment.
Coverage figures for cleaning robots assume open floor. No obstacles, no refilling, no emptying. Reality is materially lower.
Travel speed is a maximum on a clear straight path. In a space with people it is considerably slower, and correctly so.
What to request. A simulation using your actual path and payload. Most suppliers will run one during quotation, and it should be provided in writing.
How to use catalogue figures. As a way to compare two products against each other, never as a prediction of your output.
Movement and physical fit
The group determining whether a robot works in your space at all.
Threshold height it can cross. Measure the highest in your route and compare. This single check rules out many products immediately.
Minimum width required. Compare against your narrowest passage at a busy time, with people present and furniture in its normal disarray — not when the space is empty.
Turning radius. Critical in confined spaces and consistently overlooked.
Carpet and surface capability. Including pile depth.
Slope capability. Entrance ramps, basement inclines.
Obstacle behaviour. Whether it waits, routes around, or reports a fault. This affects daily experience more than any published number.
The cheapest verification. Push a trolley of comparable dimensions along the full route at a busy time. If the trolley struggles, the robot will too — with the difference that a trolley can be manoeuvred and a robot cannot.
Comparing two products fairly
A short procedure.
Write down three things you need it to do. Specifically. Then ask each supplier how their product does those three.
Ignore features you will not use. They inflate price and contribute nothing.
Normalise measurement conditions. If one quotes battery life in eco mode and another in normal mode, convert to a common basis before comparing.
Test both in your space if possible. If not, at minimum hear both operate live.
Compare what comes with them. Warranty, spares, support, content preparation. This is usually where products differ most and where they are compared least.
Compute annual cost for both. Not machine price.
Ask both the hard question. What does this product do poorly. Whoever answers specifically is usually the more reliable partner.
Frequently asked questions
Why is battery life the most misread figure?
Because it is measured in a low-power mode with the screen dim and audio off. Real operation with a bright screen, active audio and continuous movement gives substantially lower numbers.
How much rated payload is actually usable?
Roughly half as an early estimate, once the gripper and tooling mass and the centre of gravity offset are accounted for. Inertia limits are rated separately and a light bulky item can exceed them.
How should catalogue cycle times be used?
To compare two products against each other, never as a prediction of your output. Request a simulation with your actual path and payload, provided in writing.
What is the cheapest way to check physical fit?
Push a trolley of comparable dimensions along the full route at a busy time. If the trolley struggles, so will the robot — and unlike the trolley, the robot cannot be manoeuvred around obstacles.
More in Myths and questions and Robots in Vietnamese business.
