A mobile robot is not an arm on wheels

Everything that makes a robot arm project succeed is about the work envelope. Almost nothing that makes a mobile robot project succeed happens inside one.

An arm is bought on payload, reach and cycle time, and it lives in a cell you can draw a box around. A mobile robot is bought on a route, and the route runs through your entire building, which means the project is not really about the robot. It is about the building.

Start with the floor, because it is the most common reason these projects disappoint and the least likely thing to appear in a quote. Mobile robots navigate by comparing what they see to a map, and they need traction and clearance to move a load safely. Expansion joints, dock plates, drainage grates, floor slope, standing water, worn concrete and the ridge where two pours meet are all real obstacles to a machine with small wheels and a tall load. Walk the actual route before anyone quotes it, at the actual time of day it will run, and look down.

Then look at what the route passes through. Doors that need to open, fire doors that must not be propped, elevators, air curtains, and the aisle that is technically wide enough until the day a pallet is staged in it. Each of these is an integration, not a detail. A door the robot must open needs a controller and a signal, an elevator needs a far more serious one, and the aisle needs a rule that people actually follow. That last one is a management problem wearing an engineering costume, and it is the reason a cell that passed acceptance testing gets switched off in month three.

Ask early who owns traffic. One robot needs no traffic management. Six robots, two of them from a different manufacturer, sharing a corridor with three fork trucks and a shift change, need a fleet manager, and you should know before you buy whether that software is included, what it costs, whether it handles mixed fleets, and what happens when it decides two robots should wait for each other in the same doorway.

Wireless coverage is the failure nobody plans for. These machines depend on a network connection for fleet coordination and task assignment, and a warehouse or plant that has perfectly adequate coverage for a handheld scanner can have dead zones at robot height, behind racking, or in the one corner where the route turns. A site survey before purchase is cheap. Discovering the dead spot with a loaded robot stopped in an aisle is not.

Do the charging arithmetic honestly, because it silently sets your fleet size. Runtime, charge time and opportunity charging windows determine how many robots you need to cover a given number of moves, and a robot that charges for an hour is not available for that hour. If the vendor quotes runtime under ideal load on flat floor, ask what it is with your actual payload on your actual route with your actual grade, and then ask where the chargers go and who runs power to them.

Now the standard, because this is where buyers and even some sellers get it wrong. Stationary industrial robots are governed by ANSI and A3 R15.06 and ISO 10218, and collaborative applications add ISO/TS 15066. Industrial mobile robots are governed by ANSI and A3 R15.08, which is a separate standard addressing hazards the arm standards were never written for: navigation, braking distance, the behavior of a moving machine around pedestrians, and what happens when a load shifts. If somebody hands you a risk assessment for a mobile fleet that cites only R15.06, they have assessed the wrong thing. Ask which standard was applied and who signed it.

There is a strategic point buried in all of this, and it is the reason we raise mobile robots with shops that did not ask about them. An arm removes a task. A mobile robot removes travel, and travel is usually invisible in the numbers because nobody logs it. Before you can justify one, somebody has to measure how many trips happen per shift, how far, carrying what, and what the person pushing the cart would otherwise be doing. That measurement is unglamorous, it takes a week of tally marks, and it is the single highest value thing you can do before spending anything. Frequently it shows the trips are real and the payback is good. Occasionally it shows the answer is a better cart, a relocated rack, or a change to where material is staged, and you have saved yourself a fleet.

None of this argues against mobile robots. We track them, the technology is genuinely good now, and the labor they replace is the walking nobody wants to do. It argues that the diligence is different. For an arm, we ask what the part weighs. For a mobile robot, we ask to see the route.