A cobot that looks capable on a datasheet can still be the wrong choice on the production floor. The industrial cobot selection guide that matters is not a catalogue comparison. It starts with the actual part, process time, operator movement, tooling and quality requirement at the workstation.
For a repetitive screwdriving cell, a few millimetres of reach or a missing torque record can decide whether the investment pays back. For machine tending, the decisive factor may be safe access to the machine door and enough payload for the gripper, part and any extension. The right selection process turns these details into a deployable, economically justified application.
Start with the task, not the robot arm
Define the task in measurable terms before selecting a cobot model. Describe what enters the cell, what the cobot must do, what leaves it, and where human intervention remains necessary. A clear process description often exposes whether collaborative automation is appropriate at all.
Record the current cycle time, shift pattern, annual production volume, changeover frequency and defect rate. Also capture the operator’s physical actions. If the worker spends most of the cycle making judgement calls on variable parts, a cobot may not be the first improvement to make. If the work is repetitive, ergonomically demanding and governed by consistent rules, it is a stronger candidate.
The business case should include more than labour cost. Consider rework, scrap, missed torque records, inconsistent dispensing, machine idle time and the difficulty of staffing unpopular repetitive roles. A cobot does not need to replace a person to deliver value. It can free an experienced operator for inspection, setup or higher-value assembly while stabilising output.
Industrial cobot selection guide: size the real load
Payload is often misunderstood because buyers focus only on the component weight. The cobot must carry the full moving load: workpiece, gripper, adaptor plate, cables, screwdriving spindle or dispensing valve, and sometimes a camera or sensor. Add a sensible margin for variations in material, part presentation and future tooling changes.
The centre of gravity matters as much as the kilogram figure. A long gripper holding a component away from the wrist creates greater torque than the same component held close to the flange. This can reduce usable performance even when the stated payload seems sufficient. Suppliers should assess the complete tool assembly, not just the product being picked.
Reach must be checked against the full motion path. Can the arm collect a part from the infeed, access every process point, clear guarding and place the finished part without awkward joint positions? A reach that is just adequate on a layout drawing may be inadequate once cable routing, fixtures and safe approach paths are included.
For many assembly and handling applications, a compact 5 kg class cobot is sufficient. Larger systems become relevant when the part, end effector or required reach increases. FAIRINO FR-series cobots cover 3 kg to 30 kg payloads, but selecting the largest available arm without need can increase cost, footprint and cycle time. The objective is adequate capacity with controlled, repeatable movement.
Treat cycle time as a system question
A cobot’s speed specification does not equal cell output. Real cycle time includes picking, vision or sensor checks, tool changes, machine signals, door movement, verification and placement. It also includes pauses introduced by the safety design when people share the workspace.
Measure the current process over enough cycles to capture normal variation. Then separate work that the cobot can perform from work that must stay manual. In machine tending, for example, the robot may load and unload consistently, but an operator may still manage batch changes, inspect first-off parts and respond to alarms.
A cobot is often an excellent fit where cycle time is stable but not extremely short. Where every second is critical and the task has no need for human access, a dedicated high-speed automated system may be more suitable. There is no value in presenting collaboration as the answer to every application.
Select tooling and sensing early
The arm is only one part of the workstation. Gripping, process tooling and sensing determine whether the application is repeatable enough to run unattended for meaningful periods.
A simple parallel gripper may be ideal for rigid, consistently positioned parts. Vacuum is effective for flat or sealed surfaces, but needs consideration of porosity, dust, leakage and compressed-air consumption. For irregular or variable items, compliant gripping and part-location methods may be necessary. Each option affects payload, cycle time and maintenance.
Force sensing deserves similar scrutiny. It is valuable for insertion, polishing, contour following, delicate handling and contact-sensitive assembly. However, it cannot compensate for poor fixtures, uncontrolled part variation or an unstable process. The best result usually comes from combining accurate fixturing with force-controlled motion where it genuinely adds value.
For screwdriving, specify the full quality requirement: screw type, torque range, seating method, depth, traceability and reaction-force management. A system that merely drives screws is different from one that records each result and flags an exception before a defective product moves downstream.
Build safety into the layout, not the final sign-off
Collaborative operation does not mean that every installation can run without safeguarding. Risk assessment should examine the complete application, including sharp workpieces, pinch points, tool hazards, ejected parts, stored energy and interactions with nearby machinery.
The layout influences both safety and productivity. A well-positioned fixture can keep the operator and cobot working in complementary zones. A poorly arranged cell forces unnecessary stops, long arm movements and awkward replenishment. Where the process involves hazardous tooling, high forces or fast motion, additional protective measures may be required.
Plan access for loading, maintenance and recovery from faults. Operators should not need to reach around the cobot or over fixtures to replenish components. Cable routing should prevent snagging and allow inspection. These details appear minor at the concept stage, yet they often determine whether a cell remains easy to use after the first few weeks.
Check integration requirements before approving the project
The cobot must communicate reliably with the equipment around it. List every interface: machine ready signals, safety circuits, barcode readers, torque controllers, cameras, conveyors, sensors and manufacturing data systems. Clarify who owns each connection and who will test it.
Programming accessibility also matters. Internal teams should be able to recover a known fault, adjust a taught point within agreed limits and change a product programme without depending on an external specialist for every small update. That does not remove the need for disciplined change control. It gives production teams practical control over everyday operation.
Ask for evidence of the proposed sequence, not just an assurance that it is possible. A feasibility test using representative parts can reveal gripping problems, reflections affecting vision, tolerance issues and unexpected cycle-time losses. Testing is particularly worthwhile for insertion, welding, dispensing and mixed-product handling.
Calculate ROI from the constraints you actually have
A credible ROI model includes the cobot, tooling, fixtures, safety equipment, integration, installation, training and expected maintenance. It should also account for annual operating hours and the level of supervision the cell will need. A low purchase price is not automatically a low total project cost if the application requires extensive custom engineering.
Use cautious assumptions. If the line changes frequently, include changeover time. If the part supply is inconsistent, include manual replenishment and intervention. If the aim is quality improvement, estimate the cost of defects using real rework and scrap data rather than an optimistic percentage.
The strongest projects usually combine several gains: steadier throughput, fewer process deviations, reduced ergonomic strain and better traceability. A project with a modest direct labour saving can still be attractive when it removes a bottleneck that limits the rest of the line.
Use a practical selection process
Before ordering, take the proposed workstation through four decisions:
Confirm that the task is stable enough to automate and worth repeating at the expected volume.
Validate payload, centre of gravity, reach and cycle time with the intended tooling and representative parts.
Complete the risk assessment and layout concept, including operator access, nearby machinery and future maintenance.
Test the critical process step, then calculate ROI using the full installed-cell cost rather than the arm alone.
This approach avoids two costly mistakes: buying too little capability and paying for capability the application will never use. It also creates a clear basis for technical sign-off between production, engineering, quality and management.
A showroom trial or feasibility assessment is often the fastest way to move from assumptions to evidence. Bring the actual parts, fixtures and quality criteria. The most useful cobot project is not the most ambitious one on paper. It is the one your team can commission, operate and improve with confidence.