Three robot arm specs that don't trade off the way you'd guess. Two of them interact through leverage. The third barely trades off against either.
Every industrial robot arm datasheet leads with three numbers: payload (how much it can carry), reach (how far it can carry it), and repeatability (how consistently it returns to the same taught position). They sound like three independent facts you can read off a spec sheet and combine freely. They aren't. Payload and reach are physically coupled through leverage at the base joints — quote a single payload number and you've implicitly quoted it at a specific reach. Repeatability, meanwhile, is almost entirely decoupled from both — and it is routinely confused with a fourth spec, absolute accuracy, that most repetitive industrial tasks don't actually depend on.
Payload is the maximum mass the end-effector can carry while the robot still hits its rated speed, acceleration, and accuracy — exceed it and you don't get a clean failure, you get degraded accuracy, accelerated joint and gearbox wear, or in extreme cases structural damage and an arm that can't safely decelerate. Reach is the maximum distance from the base to the farthest point the end-effector can physically access — fixed by link lengths and joint travel, defining the outer boundary of the usable workspace. Repeatabilityis how tightly the robot clusters around the same programmed position across thousands of repeated cycles — a mechanical-design number, driven by backlash, joint stiffness, and encoder resolution, that has nothing to do with whether that position is the "correct" one by any outside measurement.
A robot's base and shoulder joints have a fixed maximum torque they can produce. Torque is force times lever-arm distance, so the farther the end-effector sits from the base, the less end-effector weight that same fixed torque can support at rated speed and accuracy. That's why a robot's headline payload number is quoted at a specific reference reach — often a short-to-moderate one — and why effective payload capacity tapers off as the arm extends toward its maximum reach. Manufacturers publish this as a payload-vs-reach curve (or a payload-vs-position "load chart"), not a single number, for exactly this reason.
Repeatability doesn't follow that same logic. It's a function of how much slop exists in the gear trains and joints (backlash), how stiff the structure is under load, and how finely the encoders resolve position — none of which is directly set by how much weight is on the end-effector or how far the arm is extended. A well-built arm can be both highly repeatable and payload-limited at reach; the two specs are engineered somewhat independently, even though sustained near-max payload at long reach will accelerate the wear that eventually degrades repeatability over the robot's service life.
It can't — not reliably, and not at full reach. The quoted payload figure is almost always specified at a particular reach condition, typically nearer the base where the lever arm on the shoulder and elbow joints is shortest. Extend the same end-effector mass out toward maximum reach and the torque demand on those joints rises for the identical weight, because torque scales with distance from the pivot. Push past what the joints and drives are rated for at that reach and the result isn't a clean cutoff — it's degraded path accuracy, faster-than-expected wear on gearboxes and bearings, or in a worst case an arm that can't decelerate safely and becomes structurally or dynamically unstable. Real application engineering means checking the manufacturer's payload-vs-reach curve for the actual task geometry — not just quoting the single headline number.
Repeatability only tells you how consistently the robot returns to the same position it was taught — it says nothing about whether that position matches some independently measured "true" coordinate. A robot can be extraordinarily repeatable (returning within a few hundredths of a millimeter of the same spot, cycle after cycle) while still being off by several tenths of a millimeter, or more, from where an external measurement system says it should be. That gap is absolute accuracy, and for most industrial work — pick-and-place, spot welding the same joint, dispensing at a taught point — it barely matters, because the robot was taught those positions by demonstration or offline programming referenced to its own coordinate frame, not commanded to absolute coordinates from an outside source. Absolute accuracy becomes critical mainly when a robot has to hit a position it was never taught — for example, one computed from CAD data, from a vision system, or shared with another robot or fixture that used a different reference frame — which is why accuracy-critical applications often add external calibration or vision-guided correction rather than relying on the repeatability spec alone.
Explains the three specs that headline every industrial robot arm datasheet — payload, reach, and repeatability — and why they don't trade off the way most people assume: payload and reach interact through leverage at the base joints (so a single quoted payload number applies only at a specific reach), while repeatability is a largely independent, mechanically-driven spec that is routinely confused with a fourth spec, absolute accuracy.
A robot's payload rating is the maximum end-effector mass it can carry while still meeting its rated speed, acceleration, and path accuracy. Because torque at the base and shoulder joints equals force times lever-arm distance, the farther the end-effector sits from the base, the more torque the same mass demands from those joints. Manufacturers therefore quote a headline payload figure at a specific, often short-to-moderate reference reach, and publish a payload-vs-reach curve (or a load chart plotted against position and orientation) showing how much the effective payload capacity tapers off as the arm extends toward maximum reach. Exceeding the payload actually available at a given reach doesn't fail cleanly — it shows up as degraded path accuracy, accelerated wear on gearboxes, bearings, and drives, or in extreme cases structurally unsafe deceleration.
Reach is the maximum distance from the robot's base to the farthest point its end-effector can physically access, set purely by link lengths and joint travel limits. It defines the outer boundary of the reachable workspace (often visualized as an envelope or cross-section), independent of what the arm is carrying — a robot can always physically extend to its maximum reach, but how much payload it can carry while doing so is a separate question answered by the payload-vs-reach curve above.
Repeatability measures how tightly a robot clusters around the same programmed position across many repeated cycles — a mechanical-design property governed by backlash in the gear trains, structural and joint stiffness, and encoder resolution. Absolute accuracy measures how close that position is to an independently, externally measured "true" coordinate. A robot can have excellent repeatability (a cluster spread of a few hundredths of a millimeter) while its absolute accuracy is off by several tenths of a millimeter or more, because the cluster itself can sit slightly offset from the true target due to kinematic modeling error, gravity sag, or thermal drift that repeatability testing never exposes. For most repetitive industrial work, positions are taught by demonstration or offline programming referenced to the robot's own frame, so repeatability — not absolute accuracy — is the number that actually governs task success.
Payload and reach are coupled by leverage: extending the arm farther reduces the payload it can safely carry at that extension, even though the two are reported as separate headline numbers. Repeatability, by contrast, is not a direct trade-off against either — it comes from mechanical design quality rather than from how much weight is on the end-effector or how far the arm is reaching at the moment. The most common engineering mistake is treating the single quoted payload figure as valid everywhere in the workspace, when in reality it should always be checked against the payload-vs-reach curve for the specific task geometry.
No. The headline payload figure is typically specified at a particular reference reach, usually nearer the base. As the arm extends farther out, the torque demand on the base and shoulder joints rises for the same end-effector mass, so the effective safe payload decreases. Manufacturers publish a payload-vs-reach curve (or load chart) for exactly this reason, and application engineering should check that curve for the actual task geometry rather than assuming the single quoted number holds everywhere.
It rarely fails cleanly. Typical consequences include reduced path and positioning accuracy, faster-than-expected wear on gearboxes, bearings, and drive motors, excess joint backlash developing over time, and in severe cases an arm that cannot decelerate safely, creating a structural or dynamic instability hazard.
No, and this is one of the most common confusions in robot specification. Repeatability is how consistently the robot returns to the same taught position across many cycles. Absolute accuracy is how close that position is to an independently, externally measured true coordinate. A robot can have excellent repeatability while still being measurably inaccurate in absolute terms.
Because most repetitive tasks — pick-and-place, spot welding the same joint, dispensing at a taught point — have the robot's positions taught by demonstration or offline programming referenced to the robot's own coordinate frame, not commanded from an independent absolute coordinate source. What matters is that the robot returns to that same taught point reliably, which is exactly what repeatability measures. Absolute accuracy becomes critical mainly when a robot must hit a position it was never taught, such as one computed from CAD data or a vision system.
Not directly. Repeatability is driven mainly by mechanical design factors — gear-train backlash, joint and structural stiffness, and encoder resolution — rather than by how much weight is on the end-effector or how far it is extended at a given moment. Sustained near-maximum payload at long reach can accelerate mechanical wear that degrades repeatability over the robot's service life, but that is a wear effect, not a direct spec trade-off.
Integrators plot the actual task geometry — the mass of the end-effector plus workpiece and the distance and orientation at which it must be carried — against the manufacturer's payload-vs-reach or load chart data to confirm the robot can perform the task at rated speed and accuracy throughout the required motion, rather than relying on the single headline payload figure.
Try our Robotics & Automation Studio
More calculators, simulators, and guides for this discipline.