On the plant floor, the difference between a cobot and an industrial robot usually gets reduced to one sentence: one runs behind a fence, the other does not. That shortcut hides where the two architectures actually diverge, and it leads to the wrong choice. The decision rests on required peak rate, the heaviest load, the target pallet height, the floor space you can genuinely use and the outcome of an application-specific risk assessment.
No visible fence does not mean no guarding
Collaborative robots are built to work near people, but that does not mean guarding can be dropped in every application. What governs safety is not the robot alone: it is the end effector and the load it carries. A sharp-cornered crate or a heavy bag creates risk regardless of the robot's force limits. A collaborative cell therefore still gets a risk assessment, and the result is sometimes partial guarding and sometimes a full perimeter.
Cortiva's compact palletizer is the concrete example: it is a cobot palletizing cell, and it puts the robot, its guarding, the pallet magazine and the product infeed inside a single enclosed frame. There is no separate fence line to build on site, because the guarding arrives as part of the machine. The word "cobot" therefore says nothing about the absence of guarding; it says where the guarding has been resolved.
In practice, rate is what separates the two
Collaborative operation requires the robot to slow down when a person is nearby, and that shows up directly in cycle time. Cortiva cobot palletizing cells run at up to 10 cases per minute: up to 35 kg and 2450 mm pallet height in the Standard frame, up to 50 kg and 2850 mm in the XL frame. Pallet size is up to 1200 × 1000 mm in both frames and the product is held with a vacuum gripper. These figures come from the compact palletizer entry in the product catalogue.
Where a higher rate or a different pick method is required, the industrial architecture takes over. Purpose-built palletizing cells reach up to 30 products per minute, 57 kg and 2743 mm pallet height; product can be picked by unit, by row or as a full layer, and infeed can be single or multiple. Bag palletizing works to a different range again: up to 20 bags per minute, bag weights from 5 to 50 kg and pallet heights up to 2400 mm.
What "payload" actually means
The weight limit of a cell is not the net weight printed on the case. What the robot has to carry is the product plus the gripper, and a gripper is not a light part: its frame, the vacuum generator, the valve island and the hose package hang from the same wrist. The mass difference between a single-pick arrangement and a row-pick arrangement comes straight off the available payload.
The second detail is where the weight sits. A bag whose contents move behaves nothing like a rigid case of the same weight: the centre of gravity shifts in motion and the effective load rises momentarily. Close to the limit, "how many kilograms" therefore has no answer on its own.
Footprint, layout and fitting into an existing line
The cobot cell's strongest argument is its footprint. An enclosed cell can be added to the end of an existing line without taking production down for long, which is the real advantage in plants that have no room for a full palletizing island. An industrial cell needs a larger layout, planned together with guarding, infeed conveyors and the full-pallet discharge route.
Three things are routinely left out of the layout: the operator's route for pallet changes, forklift manoeuvring space and maintenance access. All three occupy floor outside the cell and none of them appear in the machine's own dimensions. So the question is often less "which is better" than "how much floor is genuinely available".
Pallet height locks the decision early
Target pallet height is a building decision as much as a machine decision. Door height, rack opening height and forklift lift height can all pull it down. A pallet stacked higher than you can move through the plant is not reclaimed floor space; it is an output that has to be rebuilt. Measure the target height before choosing the architecture.
Day-to-day operation and flexibility
The two also differ in daily use. Cobot cells are typically programmed through a simpler interface, which suits lower-volume production with frequent product changes where operators intervene themselves. Industrial cells get their flexibility from recipe management: pallet patterns live in the control system and a new product needs no mechanical change. Which one is more flexible depends on where you need the flexibility.
Peripherals belong to the scope in both architectures. If nobody has answered where the empty pallet comes from, who places the slip sheet and how the full pallet leaves, the robot runs but an operator still waits beside it. Slip sheet and pallet feeding is not an accessory added later.
How a wrong choice comes back
- Size the peak on the average and the cobot cell cannot keep up when product arrives in bursts; the buffer fills, the conveyor stops and the operator goes back to stacking by hand.
- Write the payload without the gripper mass and a load that fits the limit on paper does not fit on site; the usual compensation is a slower cycle, and the rate target is lost.
- Skip the check against building dimensions and the machine stacks high while the pallet cannot be moved through the plant.
- Draw the industrial cell without guarding and the discharge route and the layout narrows the forklift aisle, leaving safety to a fence bolted on afterwards.
- Base the risk assessment on the word "cobot" rather than the real gripper and the real load and the outcome changes during commissioning, taking the project scope with it.
Edge cases
The decision gets hard at the extremes. Products around 35 kg sit at the ceiling of the Standard frame, where gripper mass and centre of gravity decide the outcome, while the XL frame goes to 50 kg. A target above 2450 mm rules out the Standard frame; the XL frame reaches 2850 mm and purpose-built palletizing cells reach 2743 mm, so "industrial always stacks higher" is not true. A peak above 10 cases per minute is the boundary of the cobot architecture. Pallets assembled from different products per order are a different class of problem altogether and belong under mixed palletizing.
Questions worth asking on site
- How many products arrive per minute in the peak hour, and does product flow evenly or in bursts?
- How heavy is the heaviest product, and what is the total mass once the gripper is added?
- What is the target pallet height in millimetres, and can doors, racking and forklifts handle it?
- How much clear floor can be given to the cell, and where do forklift and pedestrian routes run?
- At which point does an operator approach the cell, and how often?
- How many format changes happen per shift, and who will define the pallet pattern?
- Where do the empty pallet and slip sheet come from, and where does the full pallet go?
Common misconceptions
"Cobot means no fence." Cortiva's cobot palletizing cell is fully enclosed: the guarding arrives with the machine frame instead of being built as a fence line on site.
"A cobot cell needs no risk assessment." The assessment is not done for the robot on its own but for the gripper at the end of it and the load it carries, and the result changes with the application.
"An industrial robot is always faster." The industrial architecture can reach a higher rate, but real output also depends on the infeed, the gripper concept and the pallet change time. A badly fed industrial cell can produce fewer pallets than a well fed cobot cell.
"Operators program a cobot themselves, so no engineering is needed." The interface can be simple, but the pallet pattern, the infeed design, the gripper choice and the guarding decision remain design work.
How to reach the decision
A workable order: establish peak rate and heaviest load first, verify the target pallet height against building dimensions, then measure the floor space you can actually use, and finally have the risk assessment done against the real application. After those four inputs, the shortlist usually collapses to a single architecture. Any cobot versus industrial debate held before those numbers exist is a matter of preference, not engineering. The systems in the palletizing family separate along exactly those four inputs.