Choose an industrial robot arm by describing the heaviest complete load and every position and orientation the tool must reach, then eliminate models that cannot carry that load through the required path. A payload number and a maximum reach number are only the first screen. The load’s center of gravity and inertia, wrist orientation, access around fixtures, positional tolerance, mounting, and controller connections decide whether the arm can do the job in a real cell.
For a machine-tending or pick-and-place project, write down one task before opening a catalog: the part and gripper masses, pickup and placement coordinates, tool orientation at both ends, approach and retreat paths, takt target, available floor or overhead space, and environmental conditions. The numbers below are manufacturer specifications checked on October 14, 2026, or explicitly marked assumptions. They are examples of how to screen candidates, not a performance ranking or a verified cell design.
Start with everything the wrist carries
Count the part, gripper, fingers, adapter plate, sensors, fasteners and any cable or hose load borne by the wrist. Use the heaviest part variant and the heaviest tool configuration that will actually run. Record the combined center of gravity relative to the mounting flange and the mass moments of inertia, along with the intended speed and acceleration. A 5 kg mass close to the flange and 5 kg spread far from it impose different wrist demands. KUKA’s load-analysis description explicitly asks for mass, center of mass and inertias and evaluates static and dynamic loads at the flange and axes; a catalog payload alone cannot complete that check.
Do not casually subtract a body-mounted cable from the wrist load, or count all cable length as if it hangs from the flange. The routing and support points determine what the moving arm carries. Ask the supplier or integrator to check the actual tool drawing and worst-case part against the chosen model’s load diagram or load-analysis tool. This is also where a larger gripper, a longer finger or a second part can invalidate an apparently comfortable payload margin.
Reach the work, with the tool facing the right way
Maximum reach is a geometric extent, not a promise that every point inside a circle is usable with every tool orientation. Plot the robot base, machine opening, pickup, placement, clearance obstacles and tool center point (TCP). Check the whole path, including the approach to a chuck or tray and withdrawal with the part attached. A wrist may reach the target point but lack a valid orientation or collide with a door, guarding or fixture on the way in. Confirm joint limits and singularity or cable-wrap risks in the vendor’s simulation with the intended mounting position.
The current product pages illustrate why model suffixes matter. ABB lists the IRB 1200-7/0.7 Gen2 at 7 kg payload and 0.7 m reach, while the IRB 1200-5/0.9 Gen2 on that same page is 5 kg and 0.9 m. FANUC lists the M-10iD/12 at 12 kg and 1,441 mm, with six axes, floor/angle/inverted mounting and an R-30iB Plus controller. These are catalog facts for exact variants, not proof that any of them can execute a particular path at the desired cycle time.
| Exact manufacturer variant | Published payload | Published reach | What the specification helps screen |
|---|---|---|---|
| ABB IRB 1200-7/0.7 Gen2 | 7 kg | 0.7 m | Shorter reach, higher stated payload than the 5/0.9 Gen2 variant; still needs a load and path check. |
| ABB IRB 1200-5/0.9 Gen2 | 5 kg | 0.9 m | Extra nominal reach does not compensate for a tool-plus-part load above 5 kg. |
| FANUC M-10iD/12 | 12 kg | 1,441 mm | Its larger nominal envelope may clear a distant station; the cell layout, wrist load and path still need verification. |
The ABB page names Gen2 variants. An older IRB 1200 data sheet describes different generation and controller combinations; do not transfer its performance or option values to Gen2 without checking the exact product specification. The FANUC page does not show a publication date; its specifications were checked on October 14, 2026. Model configuration and availability must be reconfirmed with the regional supplier when quoting.
Match axes, mounting and environment to the approach
Write the required tool orientation beside each position. A catalog’s six-axis label counts controllable joints; it does not guarantee every combination of position and orientation throughout the workspace. The FANUC example has six axes, but a tilted entry into a machine still needs a feasible joint configuration and a clear approach. If the process requires only a constrained set of orientations, describe that restriction explicitly rather than requesting arbitrary wrist motion that the task never uses.
Mounting is a configuration to verify, not a free way to extend reach. The M-10iD/12 page lists floor, angle and inverted mounting; use the installation drawing, load conditions, anchoring and controller options for the chosen direction. State coolant, dust, washdown, temperature and tool-media requirements in the quote packet, and ask which robot parts and options are qualified for those conditions. A family name or an ingress-protection label does not establish compatibility with every chemical or cleaning process. Include clearance for doors, maintenance and operator access as well as the production path.
Repeatability is not the placement tolerance
Repeatability describes how closely the robot returns to a taught position under specified conditions. Absolute position accuracy describes how closely it reaches a commanded coordinate. ABB’s 2024 technical review distinguishes repeatability, absolute accuracy and path accuracy. FANUC’s M-10iD/12 product page specifies ±0.02 mm listed repeatability. That figure does not establish ±0.02 mm part placement accuracy in a customer fixture. Tool calibration, part location, gripper compliance, load, approach direction and process conditions add error.
Define the allowable error at the part or tool, not just at the robot flange. If a pin must enter a tight bore from coordinates generated offline, ask how the proposed system will establish the fixture frame and verify absolute or path accuracy. If each placement is taught and the part is mechanically located, repeatability may be more relevant. Request an acceptance test that measures the actual part in the actual fixture over the required operating conditions. A vision system or calibration option may help, but neither can be assumed from a bare arm specification.
A worked screen for one assumed machine-tending job
Suppose a shop wants to transfer a 2.8 kg part with a 1.7 kg gripper, 0.4 kg fingers and adapter, and 0.2 kg of wrist-supported sensor and cable load. The assumed moving total is 2.8 + 1.7 + 0.4 + 0.2 = 5.1 kg. This is a planning example, not a measured tool or a tested cell. It immediately rules out the 5 kg ABB IRB 1200-5/0.9 Gen2 on the stated scalar payload alone; the 7 kg ABB IRB 1200-7/0.7 Gen2 and 12 kg FANUC M-10iD/12 remain candidates on mass only.
Now assume the farthest required wrist-flange position, after accounting for the assumed tool offset, lies 0.78 m from the proposed base axis. The ABB 0.7 m variant does not pass that simple reach screen at this base position, even before tool orientation and access are checked. The FANUC 1,441 mm figure leaves it as a candidate for a detailed path study; it does not establish that the approach fits. Moving the base or changing the fixture could change the reach result, so record the layout revision used for each quote. The combined load center, inertia, wrist moments, collision-free path, cycle time and placement tolerance remain unverified for every candidate. The example deliberately stops before claiming a selection.
What to send with the quote request
Give each vendor or integrator the same packet: a dimensioned cell layout and obstacle model; pickup, placement and approach poses with TCP orientation; the part mass range; tool assembly drawing, center of gravity and inertia; duty and cycle-time target; mounting and anchoring constraints; ambient temperature, liquids or dust; required controller interfaces, I/O, vision and machine handshake; and the part-level acceptance tolerance. Ask for the exact robot variant, controller and option list, a vendor load-case report, a simulated reachable path with clearance, a cycle-time estimate with assumptions, installation needs, and a proposed on-site acceptance test. Keep robot price, tooling, guarding, integration, commissioning and support as separate quote lines so scope differences are visible.
This screen does not approve a robot cell or its safeguarding. Include the applicable risk assessment and integration work in the project; the site’s collaborative robot safety modes guide explains related safety terminology without substituting for a cell-specific assessment. If the question becomes drive sizing inside a custom mechanism, use the separate robot motor sizing guide. For this purchase decision, the next useful result is a short list with documented load and path checks against one frozen task specification.