Automatic inspection systems vs Human Visual Inspection

10/20/2020 / Author: Jorge Broto
Machine vision systems sounds like … Rocket Science, isn’t it? According to the Automated Imaging Association (AIA), Computer vision refers in broad terms to the capture and automation of image analysis with an emphasis on the image analysis function across a wide range of theoretical and practical applications.


Machine vision refers to the use of computer vision in an industrial or practical application or process where it is necessary to execute a certain function or outcome based on the image analysis done by the vision system.

In summary, we automate what we can see with our eyes and subsequently process with our brain, for those processes that are repetitive, require precision and objective criteria, among others.

What is one of the biggest concerns in auto manufacturing processes, and especially in automotive manufacturing processes today?

A key factor that influences each and every answer is ensuring the quality of the product. Do you think sales will increase if the product has inadequate quality? How to prevent defects in our product from reaching to the end user? How can we avoid unnecessary repairs during the manufacturing process?

Customer complaints, which not only have implicit costs in repairs but also in the brand image, could have been avoided with the appropriated quality controls.

In the industry in general, quality control is the process by which you ensure that a product is free of faults, problems in its functionality and any other type of tests; problems that you can imagine, applicable to your product. In automobile manufacturing, this means that vehicles pass rigorous and exhaustive tests to ensure their correct design, safety and comfort.

Regarding quality control processes automation by Machine Vision solutions, we are wondering to ourselves: why are we aiming to replace human by machines? Can machines today surpass our ability to think? To our brain?

Studies on human factors in industrial inspection date from 1950-1960, are initially related to critical inspection tasks such as maintenance in the aviation sector.

Search

  • Time available versus surface to inspect
  • Nature of the defect
  • Mix of defects
  • Probability index

External factors

  • Illumination
  • Noise,
  • Temperature

Individual factors / Decision

  • Recognition "Good Inspector"
  • Training
  • Visual fatigue

Management interaction
When the operator's decisions are contradicted by managers, or when the pressure for workload increases Interaction between operators: witness pass in shift change, loss of attention Working hours, subjectivity around the duration of shifts, work in special shifts: night shift, weekend.

Nowadayas, final inspection is still performed in most cases by visual inspection by means of trained operators. Unfortunately, according to the FMEA (Modal Analysis of Failures and Effects) inspection process guidelines, the "human visual inspection" is effective, reliable in only 80% of cases based on the observation of multiple factors such as different points of sight or operators, reduced cycle time, visual fatigue and defects not detected by the human eye, among others.

Why do we have to automate systems through artificial vision? What is artificial vision? It is not science fiction, at the end of the day, it is about doing by means of automatic inspection of images, what our eyes can observe and our brain process. The human eye commanded by its processor, the brain, is infinitely more powerful than any camera associated with an automatic processor. The capacity for autofocus and decision that we have is infinitely greater, but like any mechanism (and in this case I mean, for example, the approach that we carry out unconsciously) it entails wear and tear, it entails fatigue ...

"Ocular stress is the muscular fatigue that our eyes suffer when making an effort to focus their attention for long periods of time. This causes visual information not being processed efficiently. When we’re observing a distant object, the visual system is relatively relaxed. But when we’re looking at a close one, different mechanisms are put into operation (accommodation, convergence, pupil contraction). When it occurs for a long time, manifestations related to visual fatigue appear”

Often, there are no visual symptoms, they are only subjective discomfort and redness of the eyes. When there are symptoms in the vision, they usually appear as a transient blurring of the image, a loss of sharpness, sometimes fluctuating. The normal thing is that it remits quickly just resting.

There is no single cause for eyestrain. This fatigue is a symptom that occurs under multiple conditions, and in a more generalized way: dry eye and accommodative effort, very often a combination of both.

For those of you who haven’t had the opportunity to see a car production line in operation, can you imagine the number of vehicles that can be produced per day in a car factory? 10, 100, 1000? The cycle time, that is, the time used for the production of a vehicle is often 30 to 40 seconds. This means one car every minute!

Can you imagine a person inspecting paint defects? (defects measuring less than one millimeter, not fat chips!) On an average of 600 to 700 cars a day, 5 days a week? All the hairs on my body stand on end just thinking about it!

Let me just give you some examples of a visual inspection carried out by operators that can and perhaps could be replaced by Computer Vision based on a business case which can have from one to three years ROI depending on the number of employees to be replaced. (This doesn’t means the dismissal of these employees. There are already numerous operations that still require human intervention in the manufacturing processes but that do not cause this “inhuman” fatigue that is perhaps not comparable to those ergonomic processes in which heavy loads are avoided in repetitive operations (as in the case of exoskeletons) but which involve visual and also mental fatigue (this would perhaps lead to adifferent article).

The targets of an automatic machine vision inspection system are, among others, as follows:

  • Improve quality. Obviously this is the main goal.
  • 100% parts produced verification. to avoid defects.
  • Reduce waste and increase productivity
  • Non-contact measuring devices, we will not introduce more defects in the part to be inspected due to improper handling
  • Reliability: 100% error detection
  • Faster control: in real time.
  • Universal solution for visual inspection.
  • Traceability for warranties and claims, since we can save the images.
  • Real-time feedback to engineering for process improvement.

In most cases, both client and supplier ensure the solution ROI, having among others the following benefits:

  • Consistency: objective diagnosis vs subjective evaluation.
  • Efficiency: more verification in less time.
  • Trend analysis, for process improvement.
  • Efficiency savings in labor cost
  • Claims and guarantees reduction: no errors escape undetected
  • Brand image improvement: An intangible that weighs a lot for future sales and customer loyalty

And a quick return on investment. Think about it, it is not just a matter of business (which in itself already pays for the investment in automatic inspection) but at an “ergonomic” level and therefore with the consequent social benefit, the operator is discharged from a routine work and at the same time "stressful" work in which it does not seem that there is a physical effort but that on a psychological level it has been affecting more than what we believe in the quality controls that are carried out carried out by them.