Robot performance
The cobot study where output fell and posture improved
In a controlled study, workers made fewer products with a cobot than without one: 5.35 against 8.03. The cobot did not speed them up. It slowed them down.
What improved was ergonomics. Posture scores, measured with a standard assessment, fell by about 5% on each side, meaning a lower risk of musculoskeletal strain, even as output and collaboration time dropped.
This page traces both results to the study and treats it for what it is: one measured counterexample to the assumption that a cobot always raises output.
Data covers Controlled study of a collaborative task, productivity and posture (2023). Last reviewed by a human editor before publication.
The figures and where they come from
Each figure is rated for how safely you can cite it today. Ratings judge current usability, not whether a number was ever correct.
| Figure | What it is | Source | Citation Confidence | Notes |
|---|---|---|---|---|
| 8.03 | Products made, human-only | [A] | Medium | Mean products made in the human-only modality. The higher-output condition, without the cobot. |
| 5.35 | Products made, with cobot | [A] | Medium | Mean products made in the cobot modality, significantly fewer than human-only. The cobot lowered throughput in this task. |
| 5.0% | Posture improvement, right side | [A] | Medium | Reduction in the posture-risk score on the right side with the cobot, indicating lower musculoskeletal strain. |
| 5.7% | Posture improvement, left side | [A] | Medium | The corresponding reduction on the left side. The cobot's benefit here was ergonomic, not throughput. |
Why the numbers disagree
The result cuts against the default assumption that a cobot raises output. In this controlled task, the human-only condition produced more, 8.03 against 5.35, because the cobot paced the work and added waiting. The machine that was supposed to boost productivity reduced it here.
But the study measured two things, and they moved in opposite directions. Posture-risk scores improved by about 5% on each side, so the cobot made the work physically easier even as it made it slower. Whether that trade is worth it depends on what you are optimizing, output today or injury risk over years.
This is one task under controlled conditions, and it is offered as a counterexample, not a general law. It does not prove cobots reduce output; it proves they do not always raise it, and that productivity and ergonomics can trade against each other in ways a single headline number hides.
How to cite these figures
Use the study to puncture the assumption that a cobot automatically raises output. In this measured task, workers produced fewer units with the cobot, 5.35 against 8.03.
Carry both results together. The cobot lowered throughput and improved posture, by about 5% on each side, so cite it as a productivity-versus-ergonomics trade, not a simple win or loss.
Frame it as one controlled counterexample. It shows the outcome is not guaranteed, which is the useful thing, rather than claiming cobots generally slow work down.
Where people go wrong
Reading this as proof cobots reduce productivity. It is one task; it shows the gain is not automatic, not that a loss is the rule.
Citing the output drop without the posture gain, or the reverse. The point is the trade-off, and half of it misleads.
Generalizing a controlled-study result to a production floor. Conditions, task, and pacing all differ outside the study.
How we checked
The figures come from a peer-reviewed study of a collaborative task that measured both output and posture. We accessed it through PubMed Central and confirmed the 8.03 and 5.35 output means and the 5.0% and 5.7% posture improvements in its text.
We carried both halves of the result deliberately. A page that reported only the output drop, or only the ergonomic gain, would misrepresent a study whose whole value is that the two moved in opposite directions.
As with any single controlled study, we mark it as a counterexample rather than a general finding. It shows a cobot need not raise output, which is enough to unseat the default assumption without overclaiming the reverse.
Full source list
Primary sources, with live links. Every figure above traces to one of these.
- [A]PMC (PubMed Central)2023
"Does the introduction of a cobot change the productivity and posture of the operators in a collaborative task?", PMC (PubMed Central)
https://pmc.ncbi.nlm.nih.gov/articles/PMC10411803/
Common questions
- Do cobots always increase productivity?
- No. In a controlled study, workers made fewer products with a cobot than without one, 5.35 against 8.03. The gain is not automatic, and here output actually fell.
- If the cobot lowered output, what was the benefit?
- Ergonomics. Posture-risk scores improved by about 5% on each side, meaning lower musculoskeletal strain, even as throughput dropped. The cobot made the work easier, not faster.
- Does this prove cobots slow work down?
- No. It is one controlled task, offered as a counterexample. It shows a cobot does not always raise output, not that it generally reduces it.
- How should I weigh output against posture?
- It depends on what you are optimizing. A small ergonomic gain can pay off over years in reduced injury even if it costs some throughput today. The study shows the two can trade against each other.
More data, traced to source
- A cobot's measured productivity gain: 10%, not a multiple
Collaborative-robot marketing implies large throughput gains. A time-studied assembly cell measured the actual gain at 10%. Here is the study, and why a measured number beats a marketed one.
- Do robots boost productivity? US manufacturing productivity says not lately
Robots are sold on productivity. US manufacturing labor productivity growth went from 3.4% a year to negative over the era automation accelerated. The measured record does not show the promised surge.
- Do robots really run 24/7? The measured runtime is far lower
The pitch is that robots run around the clock. The best measured data on manufacturing machine runtime shows a median of 32% and a weighted average of 54.5%. The 24/7 figure is a ceiling, not a norm.