A Tale of Two Technicians, Two Readings, and a Single Breaker: Why a Robot Reads It the Same Way Every Time

A thermal camera measures the heat radiating off a surface, and how much of that reaches the lens depends on things that have nothing to do with how hot the equipment actually is.

• How far away you stood. Standing closer reads differently from further back.
• What angle you were at. Standing straight in front reads differently from off to the side.
• What else is in the frame? A hot surface behind the thing you're measuring changes what the camera sees.
• How much of the frame the target fills. A breaker filling the shot reads differently from the same breaker as a small part of a wider view.

Move any one of those and the number moves. Your colleague stood a foot to the left, so their number is different. That's the whole explanation.

One reading on its own is nearly useless. What tells you something is a reading compared to the last one.

A breaker at 68 °F means nothing. A breaker that was 52 °F in March and 68 °F in September means somebody should go and look at it.

But that comparison only works if both readings were taken the same way. If March was captured by one technician standing close and September by someone else standing back, the difference you're looking at might be the two technicians rather than the equipment.

So you end up with a folder of accurate readings and no ability to tell whether anything is getting worse. Which is the actual question you wanted answered.

Not the camera. The camera is the same one a technician would use.

What changes is that the robot stops at the same point every time. Same distance, same angle, same framing, on every pass, consistently. It doesn't get tired, doesn't take a shortcut on the last aisle, and it doesn't hand it over to someone else in March.

Now when you put two readings side by side, the only thing that changed between them is the equipment and its condition. That's a trend. Everything before that was just a collection of numbers.

We're not saying a robot’s readings are more accurate than a good technician's. A technician with a proper camera and the training to use it will take an excellent reading.

We're saying that sometimes they'll take an excellent reading that can't be compared to the last excellent reading, because nobody can stand in exactly the same spot twice a year apart.

Repeatability is a different thing from accuracy, and for spotting something slowly going wrong, it's the one that matters.

And the technician is still essential. They calibrate the equipment, run the checks a robot cannot, and interpret what the thermal data means and what to do about it. The robot produces comparable readings, but a person decides what they are worth.

When the robot can't get a clean read, it doesn't produce a number. It records that it couldn't, and the reason, to help address it, rather than pretending to have completed the task on that specific round.

The alternative is a reading that looks fine and isn't, sitting in your record until somebody makes a decision based on it.