Robot Repair Incident

Steve Arndt, PhD, CHFP:

So we have a robot that is malfunctioning. These were welding robots in an auto parts manufacturing plant. Maintenance comes in and begins to affect a repair. During the repair process, the individual triggers the robot to move. The robot brings a part in, and the part catches the individual, contacts his head and injures him. The robot is then stopped by an employee, but the employees don't know how to free the actual individual. So they used the controls in a manner that was not advised and ended up causing potentially more injuries. There was some dispute over the timing of the events that took place. How was the robot triggered to move? What could have triggered it to move and should there have been different design elements in place that would've prevented the incident?

Because so much time passes between the injury and a lawsuit, the plant went back to operation and over time the parts on the line changed and the robot was reprogrammed to do normal other tasks.

Steve Rundell, PhD, PE:

So we brought our scanner, we scanned everything in so we could fully capture the geometry at time, and then used the photos taken on the day of the accident to move things around and place them. We had to go through the historical data and the paperwork in this facility to figure out when they moved this component to here, and also understanding all of the built-in safety features associated with this area of the plant.

So one of those critical safety features is called a light curtain, and it's basically a series of sensors that creates a plane, like a curtain in space, that if you enter into it tells the robot arms, don't do anything, don't go in this area because there's somebody there.

Steve Arndt, PhD, CHFP:

In this case there were light curtains on this robot, there were sensors present.

Steve Rundell, PhD, PE:

Essentially what we were able to determine is that he didn't properly shut the system down. Essentially, these sensors that tell the robot arm, "Hey, you can move over here now" sense whether a part is present or not. And so it senses it because it's the sensor is close enough to the part, but over time as it's working the sensor maybe loosens gets further away. And in this case, it had gotten far enough away that it was no longer giving the right information to the robot arm to proceed. He assumes that because he broke the light curtain to get there, that the system was shut down instead of properly going into the controls and shutting the system down, basically has to over this jig, he leans over it and fixes the sensor. As soon as the sensor was fixed, the robot arm said, we're good to go. And it lowered the part onto that jig.

Steve Arndt, PhD, CHFP:

In industrial settings, anytime you perform maintenance it triggers lockout tag out, which is the control of potentially hazardous energy.Interestingly, the testimony was that this was a regular adjustment that needed to be made, and one of the arguments was because it was a regular occurrence, it was excluded from maintenance under OSHA and that regular and routine maintenance allows for you to be exposed to potential hazards. But that was a really narrow and incorrect reading of what the standard is. Just because a maintenance procedure takes place regularly, doesn't excuse the employer or the employee from following the lockout tag out rules.

Steve Rundell, PhD, PE:

There was a part two for me as the biomechanical engineer. The question was, could it cause traumatic brain injury. And traumatic brain injury that involves a pretty rapid movement of the head, such that that rapid movement can cause more massive or more dense parts of the brain to move relative to less dense parts of the brain. We were able to perform a simulation to actually look at what the forces are on the head. And essentially what we found is that because his head was so near to the component, it never experienced much movement. So there really was no opportunity for the more massive parts of the brain to move relative to the less massive parts because the head as a whole didn't move much. So, I was able to address whether or not it was consistent with closed head injury, to which our simulation showed pretty clearly that it wasn't.

Steve Arndt, PhD, CHFP:

The employer had good training programs, they efficiently had procedures in place for lockout tag out, but employees will make mistakes, employees will make errors, employees will cut corners. Can you prevent all accidents and injuries in an industrial setting? No. From the even factor side of things, is that a problem with the underlying technology, the standards that apply to the technology? Or is that an issue where the employer is in the last and best position to prevent this type of incident? And that's what we found in this case.