The other day I was driving around doing some errands in my relatively modern, high-tech car when the engine cut out while I was waiting at an intersection.

Luckily, I was able to restart the car again after having put it in neutral and turning the key fully off and back on. I missed the green light, though, and the deafening honking from cars behind me didn’t ease the stress I felt in that moment.

I took the car to the mechanic at the first opportunity. All I saw on the dashboard was a yellow engine warning light. The mechanic rang later to inform me that the car’s computer had identified the crankshaft sensor had failed, so he replaced the sensor and I was on my way.

The next day I went out again and the same thing happened. I went back to the mechanic, and he told me the sensor he put in was a loaned unit from another mechanic and he didn’t trust it. He plugged his extremely “specialised” computer analysis kit into the car and the error message came up on his laptop “Error F0000235 Crank Shaft Sensor Error – Replace Sensor.”

This was the same message as the day before.

He put in a “new” sensor from his own trusted source, and I went on my way again, this time taking a one-hour trip to a customer of mine in western Sydney.

I left my customer a couple of hours later and, at the first intersection it all happened again! I got the car restarted, but you can imagine where my attention went every time I had to wait at an intersection during my trip home—now a 1 1/2 hour-trip during peak rush hour.

Will the car cut out again? And if it does, will there be a point where it might not re-start at all? I was in luck and got home without further issues.

The car went back to the mechanic the following day.

I trust my mechanic, and I know he will eventually figure out the problem for me, maybe a new ECU “Electronic Control Unit” at a significant cost. This is just a minicomputer with locked-down software that controls many aspects of the car, including diagnostics and fault reporting.

You wonder sometimes how the computer can get the diagnostics so wrong. My old sensor might have been perfectly okay, but hey! At least I will get a new one now, right?

At a cost.

My point is, if this happens with a mass-produced, off-the-shelf car with pre-tested software coding, can you imagine the scale of issues occurring in the world of coding for automated equipment from an endless number of different vendors?

Add to the mix the evolving need for integration between all automated systems, machine to machine (M2M), Machine to Human (M2H) Manufacturing Executions Systems (MES), Enterprise Resource Planning (ERP), etc., and you might start seeing a trend. If we are told the wrong thing by the computers, we will keep replacing the sensor time and time over to no avail.

Diagnostics, reporting and automated controlled responses are software coding-related tasks based on sensory input from the surroundings. The sensory inputs are product technology-centric, while software coding is process-centric. Software coding determines what, when and where the information appears, the accuracy of it and the increasingly automated responses by other computer systems to it.

Reliability, accuracy and quality of products with sensors of all kinds such as vision inspection cameras and laser positioning devices, are where technology is improving rapidly. Sensory products have become peripheral, meaning you can find them everywhere and they all, more or less, get the job done.

In general, this is good, and yet I say we still need to do research to find the most reliable components and ensure they don’t include more software coding than what’s required.

When software coding misbehaves (which it does a lot), we begin to lose our trust again in the sensory technology gains.

Can you begin to fathom the implications of having just a few layers of software coding being off? In a market where most people are product-biased and easily forget about crucial and increasing software needs?

We must start to take software coding seriously at all levels if we are to make real progress in our automation.