Robot selection should be the result of a task and cell study. Payload is important, but cycle, reach, tooling, people, safety, presentation and lifecycle determine whether the system works.
Begin with the application, not the label
“Robot” and “cobot” are useful shorthand, but they do not define the final cell. A collaborative robot may still require guarding or other protective measures, while a conventional industrial robot may be the more appropriate platform for a fully enclosed high-rate task.
Describe what the system must do, how people interact with it and what happens during normal production, replenishment, cleaning, setting, recovery and maintenance. Platform selection follows from that operating model.
Cycle and utilisation
Start with the complete cycle: approach, grip, process, travel, release, confirmation and return. Include product presentation, machine dwell, vision, tool changes and fault recovery. A headline robot speed does not reveal the sustainable cell output.
Where the required cycle is close to the theoretical limit, model realistic acceleration, payload, path, settling and upstream variability. Add a sensible operating margin rather than designing around a perfect demonstration.
Payload, reach and inertia
Payload includes the tool, product, adapters, services and any change mechanism. Reach must cover the real path while maintaining suitable joint positions and clearance. Large or offset loads can be limited by inertia and moment before nominal payload is reached.
A good concept plots the complete envelope and checks the difficult poses: low pick points, deep machine access, rotated tools and service routing.
Human interaction and safety
Decide why people need to be near the task. Is proximity essential to the process, useful only for replenishment, or simply inherited from the current manual method? The answer affects layout, safeguarding and economics.
Collaborative operation is an application-level decision that requires assessment of the robot, tool, workpiece, speed, contact possibilities and operating modes. A product marketed as collaborative does not remove the need for competent risk assessment and validation.
Tooling and part presentation
End-of-arm tooling can dominate reliability. Surface finish, porosity, flexibility, tolerance, contamination and orientation all affect the grip method. The feeding or fixture concept must present the part within the window the tool can tolerate.
Trial representative parts early when the grip is uncertain. A robot cannot compensate indefinitely for inconsistent presentation or an unstable pack.
Compare two complete cell concepts before choosing
Make the comparison fair
Give both concepts the same part range, accepted output, staffing assumption and replenishment schedule. Compare a guarded industrial robot concept with a collaborative concept using the actual operating restrictions each would require. A smaller robot price or a shorter empty motion is not an equivalent production comparison.
Model the interruption that occurs most often
Suppose an illustrative task takes 15 seconds including process dwell: its uninterrupted ceiling is four parts per minute. If material replenishment stops the cell for one minute in every five, available production time falls to four minutes out of five before other losses. The resulting planning ceiling is 3.2 parts per minute. Measure the real sequence and avoid applying the same loss twice if the quoted cycle already includes it.
Ask what human proximity is achieving
Separate shared work on the same part from occasional loading, visual checks and maintenance access. Changing the feeder, fixture or replenishment position may change whether a person needs to enter the robot envelope at all. Safeguarding still requires application-specific assessment of the tool, workpiece, motion and foreseeable interventions; the platform label does not settle that decision.
Use a small set of decision gates
Before commitment, seek a credible full-cycle model, a representative gripping or process trial where needed, a reviewed access concept and a defined recovery sequence. Record what would make either proposal unsuitable: insufficient payload margin, unreachable service access, unstable part presentation or unacceptable interruption losses. Continue with machine safety by design and component presentation planning.
Whole-life economics
Compare the complete cell: robot, controller, tooling, base, guarding, sensors, conveyors, integration, programming, training, maintenance and changeover. Include the cost of lost output if the concept has insufficient cycle margin or is difficult to recover.
The right platform is the one that meets the production objective with an acceptable risk, operating model and lifecycle—not the one with the most attractive standalone specification.
- Define the complete task and operating modes first.
- Model sustainable cell cycle, not headline robot speed.
- Include tooling and product in payload and reach decisions.
- Treat collaboration as an application assessment, not a product label.
- Compare whole-cell economics and recovery, not robot price alone.