Machine Vision and the Auto Industry

The use of machine vision has been implemented in many industries, and it’s now slowly, but surely, being integrated into one of the most mature production lines: automobile assembly lines.

Machine vision first debuted in the 1980s, where there was great interest in the manufacturing sector in automating many of the industry’s production lines. This futuristic technology touted the ability to emulate human vision, thus giving sight to previously “blind” robots, removing the need for human observers to guide these robots. However, this technology was still immature; being prone to errors, having exorbitantly high prices and the hassle that comes with installation of machine vision into production lines all deterred potential customers from purchasing this product. For the automotive industry, where reliability, cost and efficiency are paramount, the machine vision remained adamantly unexplored.Machine Vision

Fast forward 30 years – the technology for machine vision has matured. Advances such as laser triangulation and CCD cameras all play a part in elevating the status of machine visions from merely “plausible” to “effective”. In addition, with these advances, the once bulky units that cost upwards of $40,000 now have been compacted, and they only cost from $5,000 to $20,000. With their ability now expanded to improved sight, instantaneous feedback and better-equipped sensors, the automotive industry stirred. Now, many companies such as Ford and General Motors have started to implement machine vision for quality control and robotic guidance in assembly lines, and this trend is set to go upwards – machine vision usage by the automobile sector is set to grow by 20% over the next 5 years.

How the automotive industry implements machine vision

Robotic Guidance

As mentioned, the automotive industry uses machine vision for two main purposes: to inspect the quality of the sundry items comprising the vehicle, ensuring that they pass muster, and to provide guidance to otherwise “blind” robots in the assembly line, such that the margin of error is further reduced

In one instance of robotic guidance by machine vision, robots in the assembly line are given feedback when the robots install components into the mainframe of the automobile. For example, when the door frames of the car are installed, the machine vision takes a pre-determined reference point, and instructs the robot to install the door frame based on the reference point. This has helped to reduce the margin of error to almost 0.1mm. In contrast, the probability window of the installed window will be about 3mm from the preset “perfect” position with the blind robots – this will result in higher chances of defective vehicles going onto the roads, essentially putting the drivers on the roads in danger of an accident due to a failed component, tarnishing the automotive company’s name. One such example is the 2009 – 2010 recall by Toyota due to unintended acceleration by the vehicle, causing Toyota to recall more than 2 million units across the globe.

Quality Control

Machine vision also helps by giving live video feeds and statistics to human quality controllers while the cars are being assembled. In the past, the automobile must be assembled completely before the quality controllers can assess the automobile’s functionality, verify that the assembly of the vehicle is within tolerance and test the safety aspects of the car. By the time a problem is detected, thousands of the automobile may have already made it to the market, or even to the hands of consumers.

Auto Assembly LineFor example, the machine vision will monitor the “gap” and “flush” of two welded metal sheets – gap meaning the distance between the two metal sheets, and flush referring to the vertical alignment of the two metal plates. If the gap and/or flush is outside the standards set, the machine vision will feedback to the quality controllers, and they will be able to take the defective vehicle out of the production line to manually resolve the issue. This process prevents these defects from ever leaving the manufacturing plant.

With machine vision, quality control has also been improved – machine vision is able to inspect 100% of the parts in the production line, compared to the previous method of randomly choosing a certain percentage of the automobile part, and using them to test if the entire batch of components can be used for the production line. The use of laser triangulation by machine vision, for example, can provide vital feedback on the shape and size of brake pads, and they will weed out the brake plates that do not meet the required geometric specifications.

Removal of human contact

As an added bonus, machine vision has also removed human contact with potentially hazardous situations. For example, in brake pad tests, the brake pads can reach temperatures of up to 90 degree Celsius. Such extreme temperatures will be dangerous for the engineers if they were to come in close contact with these searing-hot brake pads. However, if they were to wait for the brake pads to cool before analyzing and detecting the potential defects in the brake plates, the results will be erroneous – there are flaws that can only be detected when the brake plates are still hot from friction. Machine vision solves these problems by removing the need for human contact in these situations. They are able to provide accurate feedback on each brake pad being tested, and single out the defective ones based on the standards set.

Impacts on the Auto Insurance Industry

The use of machine vision promises more reliable automobiles on the roads – with every single component within the vehicle being tested for quality at every stage of production, the amount of defective vehicles on the roads will definitely decline drastically. This decrease will correlate to higher safety standards for all road users, thus potentially decreasing the road accidents caused by these defective units. A fall in accidents will mean a drop in claims for insurers. This may result in a decision by the auto insurance companies to drop the auto insurance premiums for road users.

However, this is only a hypothesis – it is naïve to think that machine vision itself will cause a drop in auto insurance premiums. The premium rates are determined by many factors, such as the statistics on theft of a certain model, safety standards and the number of accidents involved with a particular model. It is unfitting to pin the implementation of machine vision as the sole reason why auto insurance premiums could fall in the future.

Moreover, precedents have shown that defective units might not be reason enough for increases in auto insurance premiums. In a recent statement released by various auto insurance companies, they have stated that auto insurance premiums will not increase for owners of the defective Toyota units. This may be because Toyota is reimbursing the auto insurance companies for each claim made because of their defective units. This example has shown that the presence of defective units on the roads, and the subsequent claims by their owners for the possible accidents caused by the defects will not directly cause an increase in auto insurance premiums as long as the automotive company takes responsibility.

The rise of machine vision in the automotive industry have made assembly lines safer and more efficient – the removal human contact from potentially hazardous steps in the assembly line and of time-intensive quality checks done by humans have contributed to these two improvements. Although this may collectively bring down the cost of production and subsequent the price tag of automobiles in the market, this should not directly correlate with a drop in auto insurance premiums. It might be a contributing factor, but even then, it will be just a small influence to auto insurance rates.

Be the first to write a comment.

Your feedback