When a robot stops for no reason, the body notices first

Picture a workshop floor. A small self-driving vehicle, an AGV or automated guided vehicle, rolls up to a deburring station, does its job and rolls back. An operator works alongside it. Now and then the robot stops. Sometimes the reason is obvious: a box has ended up in its path. The operator moves the box, the robot carries on. But sometimes the robot stops on a completely clear floor, with nothing in the way and no explanation at all.

That second kind of stop is the one we were interested in. And it turns out that the difference between a stop you can explain and a stop you can’t changes almost everything about how a person behaves next. Even when they’ll tell you, quite sincerely, that it didn’t bother them.

The setup

We ran the study on a real workshop floor at IDEKO in Elgoibar, not in a quiet lab. Seventeen operators each experienced two situations, several times over. In the first, the robot stopped for a visible obstacle. In the second, we quietly triggered a stop ourselves at a point where the path was clear – no obstacle, no reason the operator could see. The recovery was the same either way: press resume, carry on. The only thing we changed was whether the stop made sense.

While people worked, we recorded three things at once: how long each stop took to resolve, where their eyes went (using eye-tracking) and how their bodies moved (using camera-based motion tracking). Afterwards, we asked them how the experience had felt.

What the body says

The unexplained stop slowed everyone down. It took noticeably longer to deal with – a median of around six seconds against four and a half – even though the physical action, pressing resume, was identical.

The eyes told the clearest story. When the stop had a visible cause, people’s gaze roamed: quick, wide sweeps across the scene, taking everything in. When the stop made no sense, the gaze did the opposite – it settled. Fixations grew longer, the eyes lingered, the visual field narrowed. This is what someone looks like when they stop scanning their surroundings and lock onto one spot, trying to work out what on earth just happened.

The body did something you might not expect. Faced with the unexplained stop, people moved more, not less. Hips and shoulders sped up, movement paths lengthened and the time spent standing still collapsed. Posture became less stable and more awkward, with a measurable rise in ergonomic strain. One measure of postural change moved in the same direction for every single participant, without exception. So the hesitation didn’t show up as freezing or as jerky, uncertain movement – it showed up as restlessness. The eyes fix; the body fidgets; the task drags.

What people say and what they do

Here is the part that makes the study worth talking about. When we asked people how it felt, only one thing shifted: their sense of how clear the situation was. That dropped sharply. But their trust in the robot held steady. So did their sense of being in control and their reported effort – they didn’t feel more tired or more stressed.

In other words, people noticed that they didn’t understand why the robot had stopped and then carried on reporting that they trusted it, felt in control and weren’t bothered. Their words said they were fine. Their eyes and their bodies said something else entirely.

That gap is the finding. And the two sides aren’t unrelated: the drop in clarity was the one reported measure that tracked closely with the behaviour we could actually see. The people who felt the biggest loss of clarity were also the ones whose eyes locked hardest and whose bodies moved most.

Why it matters

As robots move onto shop floors and start sharing space with people, we tend to worry about whether they’re safe and whether they work. This suggests a quieter factor sits alongside those: whether their behaviour is legible. A robot stopping isn’t the problem. A robot stopping without a visible reason is what unsettles the person beside it – even a person who would never say so.

The encouraging implication is how small the fix might be. The stop itself wasn’t frightening. What was missing was the “why”. A simple, honest cue – a light, a message, a word – telling an operator what a robot is doing and why could close much of that gap.

The honest fine print

This was a pilot, with a small group of participants and all the messiness of a working industrial environment rather than a controlled lab. A few of the results need a longer, better-balanced study before we’d lean on them. But a solid core came through clearly: the extra time, the ergonomic strain, the restlessness in trunk and neck, the change in how people look. Those are the threads we’ll carry into the next phase and, in time, into a full conference paper.

For now, the short version is this. Ask people whether an unexplained robot stop troubles them and they’ll shrug. Watch them and you’ll see it does.