Cobot Payload: Count the Tool, the Services, and the Reach
Direct answer: A collaborative robot’s rated payload covers the end effector, its mounting plate, the services the arm carries and the part — not the part alone. Hose bundles and cabling sit inside that budget, and the centre of gravity matters as much as mass. Reach belongs in the same calculation, because the further the arm extends the fewer usable postures remain. Size payload and reach together, and revisit the risk assessment when the tool changes.
Who this is for: This guide is written for automation engineers sizing a collaborative robot for a real application, and for buyers comparing rated payload figures between models.
Scope: It covers what a rated payload has to cover, why services and reach belong in the same calculation, and when a tool change requires the risk assessment to be revisited. It does not cover process parameter development, gripper mechanical design, or programming technique.
Cobot payload and dynamic inertia: what the rating covers
The number on the datasheet is a rating for everything mounted beyond the flange, not a part-weight allowance. That includes the end effector itself, any adapter or quick-change plate, sensors mounted on the tool, and the part being held. In the reference footage for this guide the arm carries an end tool with a hose bundle running along it — the flange is not empty, and it never is in a real application.
EVST sizes on the assembly beyond the flange, not on the part. Two figures matter, not one. The mass of the assembly is the obvious one. The centre of gravity of that assembly relative to the flange is the one that is more often left out, and it changes the effective load significantly: the same mass mounted further from the flange loads the wrist joints more. Manufacturers publish payload-versus-centre-of-gravity information for this reason, and sizing that ignores it can be optimistic by a wide margin.
Inertia is the third property. A wide or long tool has a larger moment of inertia even when its mass is modest, and that inertia is what the joints have to accelerate and decelerate. It is why two tools of the same weight can behave very differently at the same programmed speed.
Services are payload too
Anything the arm carries along its structure counts. Hose bundles for pneumatics, material supply, or extraction; power and signal cable for the tool; a quick-change coupling; a camera bracket. Individually they look small. Together they routinely consume a meaningful fraction of a small cobot’s rating, and unlike the part they are present in every pose.
Routing matters as much as mass. A service package that is clamped close to the arm behaves differently from one that hangs and swings, and a bundle that resists motion adds a load the sizing calculation did not include. Poor routing shows up first as reduced usable speed and later as premature wear on the bundle itself.
EVST lists the service package as a line item in the payload budget. It is worth listing services explicitly in the sizing document rather than folding them into a round-number margin. When a tool is changed later, that list is what tells you whether the change is inside the original envelope. Anyone comparing collaborative robot dispensing coordinate and path behaviour across applications runs into the same accounting.
Reach and posture are one problem
Reach is usually quoted as a radius, which suggests a sphere of equally usable space. In practice the usable region is smaller and shaped by joint limits, singularities, and self-collision. Near the outer edge of the envelope the arm is close to extended, fewer joint configurations reach the same point, and the ones that remain may present the tool at an unusable angle.
In the reference footage the arm works across a large housing on an open bench, and the posture visibly changes as it moves to the far side of the part. That is the ordinary behaviour, and it is why a reach check against a point cloud is not sufficient. The check has to be against the tool pose required at each point, with the real tool geometry in the model.
EVST checks both in one calculation. Payload and reach then interact. Some collaborative robots derate allowable speed or acceleration in extended postures, and a load that is comfortable close to the base can be marginal at full extension. Treating them as one calculation rather than two independent checks is what prevents a cell that works in the demonstration and struggles at the corner of the fixture.
Decision table: application evidence mapped to sizing conclusions
The last row is the one that saves money later. A cell sized exactly to today’s tool will need re-engineering the first time the application changes.
| Evidence from the application | Sizing conclusion it supports | Evidence you still owe |
|---|---|---|
| Heavy tool, light part | Size on the assembly, not the part; check centre of gravity carefully | Measured tool mass and centre of gravity, wrist load at the worst pose |
| Long or offset tool | Expect derating; check moment of inertia, not only mass | Inertia figures for the tool, achievable speed with that inertia |
| Bulky service package carried on the arm | Include services in the payload budget and fix the routing | Service mass, routing drawing, behaviour at full travel |
| Work spread to the edge of the envelope | Check pose feasibility at the extremes, not just reachability | Reachable poses at the far points, speed permitted in extended postures |
| People share the space during operation | Task-based risk assessment governs the speed and force regime | Risk assessment result, validated protective measures for this task and tool |
| Tool expected to change during the cell’s life | Reserve budget and record the original assumptions | Payload and inertia envelope the design allows, re-validation rule |
Why the datasheet number is a boundary, not a promise
Robot performance figures are established under defined conditions. According to ISO 9283:1998, performance criteria and related test methods for manipulating industrial robots include the test conditions — load and velocity among them — under which pose accuracy and repeatability are determined. A figure quoted outside those conditions is not a statement about your application.
The same applies to payload. A rated payload assumes an assumed centre of gravity and inertia; exceed either and the rating no longer describes the arm’s behaviour, even if the mass is within limit. This is a normal engineering boundary rather than a defect, and reading it that way makes model comparison much easier.
This guide makes no claim about any specific robot’s accuracy, cycle, or capability. The footage behind it shows a collaborative arm carrying a tool and services across a part; it does not demonstrate a payload figure, and none is asserted.
Changing the tool changes the risk assessment
Collaborative operation is a property of the application, not of the robot. According to ISO/TS 15066:2016, the considerations for collaborative robots apply to collaborative operation itself, and according to ISO 10218-2:2025, safety requirements apply to the industrial robot application and cell. Both are framed around the task, which means a new end effector is a new task.
A heavier tool, a sharper tool, a hotter tool, or a faster motion changes the hazard picture even if the arm is unchanged. According to ISO 12100:2010, the general principles for design cover risk assessment and risk reduction, and re-running that assessment after a tool change is the normal expectation rather than an unusual burden.
It is worth writing the trigger into the cell documentation: which changes require a re-assessment, who signs it, and what has to be re-validated. Without that, a tool swap made for good production reasons quietly invalidates the safety file.
Mounting and the workstation around the arm
Where the arm is mounted determines how much of its envelope is usable. A bench-mounted arm loses the region occupied by the bench; a pedestal raises the working height and may recover it. Mounting stiffness matters too, because a compliant mount lets tool inertia move the base, which shows up as path deviation rather than as an obvious fault.
The bench itself is part of the design. Part location, the operator’s approach, where fixtures sit, and where the service bundle is anchored all influence the postures the arm has to hold. Working these out on the layout is far cheaper than discovering them during teaching.
For a plant standardising several such stations, mounting, service routing, and tool interface are usually specified once and reused, which is where wider robotic factory integration pays off.
What to send for a cobot sizing check
Send the end effector weight including its mounting plate and the centre of gravity from the flange, the heaviest part it will hold, the service package the arm carries and how it is routed, the working envelope including the furthest and lowest points the tool must reach, and the required motion speed. Say whether people share the space during operation, because that changes the regime rather than the arithmetic.
What comes back should be a payload budget rather than a model name: how much of the rating the tool and services consume, what is left for the part, and where in the envelope the margin gets thin. That evidence is what an acceptance discussion needs, and it is what an application review should hand back. If you are still choosing a hardware family at that point, the robot product range is a reasonable starting reference.
Frequently asked questions
Does rated payload include the gripper?
Yes. The rating covers everything beyond the flange — end effector, mounting plate, tool-mounted sensors, and the part. Sizing against part weight alone is the most common cause of an undersized arm.
Why does centre of gravity matter so much?
Because the same mass mounted further from the flange loads the wrist joints more. Manufacturers publish payload against centre-of-gravity offset for exactly this reason, and ignoring it makes the sizing optimistic.
Do hoses and cables really count?
They do, and they are present in every pose rather than only when a part is held. List them explicitly in the sizing document, and fix the routing, because a bundle that resists motion adds load the calculation did not include.
Do we need a new risk assessment after changing the tool?
Normally yes. Collaborative operation is a property of the application, and a new end effector is a new task. ISO/TS 15066:2016 and ISO 10218-2:2025 are framed around the application, so a tool change is a trigger to re-run the assessment.
Project inputs for an application review
Send the following and payload and reach can be checked together against the real application:
- the end effector weight including the mounting plate, and its centre of gravity from the flange
- the heaviest part the tool will hold, and whether anything else is carried with it
- the service package the arm has to carry: hoses, cables, and where they are routed
- the working envelope, including the furthest and lowest points the tool has to reach
- the required motion speed and whether people share the space during operation
Send the end effector weight with its mounting plate and centre of gravity, the heaviest part it will hold, the service package the arm carries and its routing, the working envelope including the furthest and lowest reach points, and the required speed. Those project inputs are enough for an application review that produces a payload budget, shows where in the envelope the margin gets thin, and states the acceptance evidence the sizing still owes. Related reading: workstation configuration reference, wider robotic factory integration, robot product range.