Robotics in Manufacturing

Workplace safety

What robot fatality case files actually record

In one documented case, a worker was crushed by a spot-welding robot that was not locked out. In another, a mold-setter was struck in the head by a cycling single-side gantry robot.

There is no reliable aggregate count of robot deaths that traces cleanly, but the individual investigation files do exist, and read together they point to one recurring failure: the robot was energized and able to move when a person was inside its reach.

This page traces two documented cases to their source records and draws out the shared lesson, without inflating them into a statistic they cannot support.

Data covers Individual documented robot-fatality investigations (OSHA and state FACE). Published 2026-07-20. Last reviewed 2026-07-20. Last updated 2026-07-20. Edited by Mike Ramsey / Reliable Media.

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.

FigureWhat it isSourceCitation ConfidenceNotes
not locked outOSHA case, cause[A]HighThe OSHA report states the spot-welding robot was not locked out and activated as the worker reached in. A lockout failure, not a freak event.
crushed by spot-welding robotOSHA case, event[A]HighThe recorded event: a worker crushed by a spot-welding robot while placing tips into a cartridge.
struck by a gantry robotFACE case, event[B]HighThe MIFACE case: a mold-setter struck in the head by a cycling single-side gantry robot during work at an injection-molding machine.
January 19, 2001FACE case, date[B]MediumThe date of the FACE-investigated fatality, a documented record with a full investigation narrative.

Why the numbers disagree

The aggregate robot death toll cannot be sourced cleanly, but that does not mean nothing is documented. Individual investigation files exist: OSHA accident reports and NIOSH-family FACE investigations reconstruct specific fatalities in detail. They are case records, not a count, and they are useful for exactly what they are.

Read together, the cases point to a pattern rather than a statistic. In the OSHA case the robot was not locked out and activated as the worker reached in; in the FACE case a gantry robot was cycling as a worker was struck. The recurring element is an energized robot able to move while a person was within its reach, which is the failure lockout and guarding exist to prevent.

The honest limit is that two cases are examples, not evidence of frequency. They show how robot fatalities happen and what failed, and they should be used that way, not summed or extrapolated into a rate the records cannot support.

How to cite these figures

Use the cases to show how robot fatalities occur and what fails, citing the specific OSHA and FACE records rather than an aggregate number.

Draw the shared lesson: in the documented cases the robot was energized and able to move while a worker was in reach, which is a lockout and guarding failure.

Do not turn the case files into a count. They are examples of mechanism, not evidence of how often robot deaths happen.

Where people go wrong

Summing or extrapolating individual cases into a rate. The files document specific events, not frequency.

Treating the cases as freak accidents. Both trace to an energized robot and a missing lockout, the standard robot-cell failure.

Citing the mechanism as if it were the toll. How robot fatalities happen is documented; how many occur is not cleanly sourced.

How we checked

The cases come from primary investigation records: an OSHA accident report and a Michigan FACE investigation, the latter part of the NIOSH Fatality Assessment and Control Evaluation program, hosted by Michigan State University. We retrieved both and confirmed the event descriptions and the lockout finding in their text.

We deliberately use case files rather than an aggregate, because the aggregate robot death toll does not trace cleanly while these individual records do. The value is the documented mechanism, not a count.

We are explicit that two cases are examples, not a statistic, and we draw only the lesson the records support: the robot was live when it should have been locked out.

Full source list

Primary sources, with live links. Every figure above traces to one of these.

  1. [A]U.S. Occupational Safety and Health AdministrationSeptember 2022

    OSHA Accident Report 149420.015, "Employee is killed when crushed by spot welding robot"

    https://www.osha.gov/ords/imis/accidentsearch.accident_detail?id=149420.015
  2. [B]Michigan State University, Occupational and Environmental Medicine (MIFACE)2001

    Michigan FACE (MIFACE) Investigation 01MI002, gantry robot fatality (NIOSH-family program, hosted by MSU)

    https://oem.msu.edu/images/MiFACE/01MI002v1.pdf

Common questions

Are there documented robot fatalities?
Yes. Individual OSHA accident reports and NIOSH-family FACE investigations reconstruct specific robot fatalities in detail, even though a reliable aggregate count does not trace cleanly.
What do the case files have in common?
An energized robot able to move while a worker was within reach. In one OSHA case the spot-welding robot was not locked out; in a FACE case a gantry robot was cycling as a worker was struck.
How many robot deaths do these represent?
They are individual cases, not a count. They document how robot fatalities happen and what failed, and should not be summed or extrapolated into a rate.
What is the safety lesson?
Lockout and guarding. The documented cases trace to a robot that was live and able to move when a person was in its reach, which is exactly what lockout and interlocks exist to prevent.

More data, traced to source