10 Examples of Using Machine Vision in Manufacturing
There are many applications for machine learning, whether it be processing social media traffic and trying to surface actionable insights or targeting consumers based on past purchases. In this article aimed at those interested in artificial intelligence, we look at 10 examples of machine vision in manufacturing which include the following:
- Predictive Maintenance
- Package Inspection
- Reading Barcodes
- Product and Components Assembly
- Defect Reduction
- 3d Vision Inspection?
- Improving Safety
- Track and Trace
- Plain Text Reading and Handwriting Analysis
- AI and Deep Learning – Landing.AI
A business that depends on physical components to manufacture products or help provide services often needs to undertake maintenance on machinery or equipment or in the worst-case scenario, machinery can break or components can become faulty bringing product to a standstill.
Predictive Maintenance is the process of using machine learning and IoT devices to monitor data on machinery and components, often using sensors, to collect data points and identify signals or take corrective actions before assets or components break down.
Consider that just one minute of downtime in an automotive factory can cost as much as $20,000 on high-profit vehicles. Its challenges like machine vision can help business keep on top of, for example, a software program called ZDT (Zero Down Time), developed by FANUC, collects images from cameras attached to robots, these images and accompanying metadata are then sent to the cloud for processing and helps to identify potential problems before they arise.
During an 18-month pilot, the solution was deployed to 7,000 robots in 38 automotive factories across six contents and detect and prevented 72 component failures!
It is critical for pharmaceutical companies to count tablets or capsules before placing them into containers. To solve this problem, Pharma Packaging Systems, who are based in England, has developed a solution that can be deployed to existing production lines or even ran as a standalone unit.
A key feature of the solution involves using computer vision to check for broken or partially formed tablets. As tablets make their way through the production line, pictures are taken and transferred to a dedicated PC that then processes the images using software which then runs further analysis to check if the tablets are the right color, length, width, and whole.
The PC based Vision Inspection system is also implemented to a PC that performs the counting function and if a tablet is deemed as defective, this information is logged which then sends a signal to the counting functioning, and by the time the bottle of containers reaches the end of production line, containers that have defective tablets are then rejected, thereby removing the possibility of shipping defective medical tablets.
You can read more about this solution here.
Reading, identifying and processing hundreds and thousands of barcodes per day is no easy task and something that humans simply cannot do at scale.
For example, cell phones and mobile devices require smaller and smaller printed circuit boards (or PCBs). As manufacturers are pressured to produce higher volumes of PCBs for the ever-growing tech market, they are looking towards a process known as “panelization”. In this process, a number of identical circuit boards are printed onto a large panel, each circuit is then separated by the machine for final testing, in order to inspect these boards, however, a machine vision-based solution called PanelScan was developed to read the barcodes – which are the unique identifiers of each circuit that is present on the PCN panel.
Historically a human applied this task by using a handheld barcode scanner, naturally, this was time-consuming and open to human error. By implementing a machine vision-based solution, PCB manufacturers can drive business savings.
You can learn more about how manufacturers are using machine vision to process barcodes here.
Product and Component Assembly
High performant manufacturing plants need to ensure products and components that fall off the production line adhere to quality, safety and production guidelines. It‘s with this in mind that Acquire Automation has developed a suite of solutions that help businesses ensure their product and component assembly standards are being enforced.
For example, one of their solutions implements machine vision that allows manufacturers to inspect bottles in a full 360-degree view to ensure that products are placed in the correct packaging and is also able to inspect other critical attributes of packaged products such as:
- Cap closure/seal
- Print quality and much more!
All of this helps increase the throughput of the production line whilst at the same time reducing the number of product recalls and increasing productivity and ultimately, keeps consumers happy!
You can read more about some of the other machine vision solutions at their site here.
Understandably, if you run a manufacturing line, you want to produce components or products that are free of defects! Machine vision is a technology that can help businesses achieve this.
That said, machine vision inspection systems can vary widely in terms of their implementation, some require an operator whereas more complex vision-based solutions do not need an operator.
A firm named Shelton has a surface inspection system called WebSPECTOR that identifies defects and stores images and accompanying metadata related to the image. As items fall through the production line, defects get classified according to their type and are assigned an accompanying grade.
Doing this allows manufacturers to differentiate between different types of defect who may then wish to only halt the production line when X number of Y types of defect has occurred.
Another one of Shelton’s machine vision-based technologies called WebSpector which leverages imaging software and state of the art cameras could improve the productivity of a fabric producer by 50%! You can read more about this story here.
3d Vision Inspection
Machine vision can play a massive role in the motoring sector. One report suggests that the overall machine vision market could be worth up to $14.43 billion by 2022!
A machine vision inspection system that contains a Dalsa Genie Nano camera is being used in a production line to undertake tasks that humans can sometimes struggle with. In this use case, the system uses high-resolution images to build up a full 3d model of components and their connector pins.
As components pass through the manufacturing plant, the machine vision system takes multiple scans of images from different angles to produce a 3d model, these images, when combined, allow the system to identify if connector pins on circuitry are faulty which could have disastrous effects later down the production line.
3d vision inspection has many applications but one of the most common use cases for the technology is in the production of automobiles.
With electrical faults accounting for a lot of automobile faults these days, being able to perform 3d scans of connector pins can help car manufacturers drive cost savings, reduce the chance of shipping faulty electrical components and help improve driver safety.
You can learn more about 3d inspection and this use case here.
The applications of machine vision aren‘t just restricted to productions lines in manufacturing plants. For example, Komatsu Ltd, who is a leading manufacturer of mining and construction equipment based in the UK, recently announced plans to partner with NVIDIA to integrate NVIDIA‘s set of “cloud to edge” technologies. The main driver for this was to improve site management services, safety, and efficiency.
Hire expert developers for your next project
1,200 top developers
us over the last 3 years
The partnership integrates the NVIDIA Jetson AI platform into machinery often used with drilling, excavation, and mining. A combination of real-time cameras and video analytics allows the equipment to run with greater efficiency and improved safety.
The idea is also to also apply deep learning-based artificial intelligence to track people and predict the movement of equipment to help avoid dangerous interactions thereby improving safety.
With as much as ten thousand injuries occurring in the US each year on construction sites that are associated with vehicles and machinery, solutions like this will be welcomed by firms.
You can find read some more about this story here.
Track and Trace
Pharmaceutical firms are naturally under stringent rules and regulations to ensure their products can be tracked and traced from the production line to the end patient.
To help achieve this, cartons can be printed with details that include but are not limited to, serial numbers, expiration dates, manufacturing dates. A globally unique identifier, sometimes known as a GTIN (Global Trade Item Number) is often used to allow packages to be tracked worldwide.
Manufacturing systems can autogenerate these identifiers in a master database which are then used later in the production process and sprayed onto containers and the next step of the production process can be performed, which often is the verification of the information that was just sprayed onto the carton on the packaging.
But how is this done?
You guessed it, Machine vision!
One solution developed by German firm ISW employs a solution like this that involves, amongst other things, high-tech cameras that can read data from labels – as well as perform optical character recognition (OCR) to read the printed text.
When the printed text has been read, the system can check against the master database and validate if the system printing labeling matches the data stored in the master database. If any printed codes are unreadable or don‘t match existing codes in the master database, then packages or cartons can be rejected.
You can find out more about the technology, components, and solution here.
Plain Text Reading and Handwriting Analysis
Optical character recognition is nothing new, it‘s been around for quite a while now in the computing world. That said, getting the machine to detect and extract handwritten text from notes, letters, etc. which contain images is a completely different thing altogether.
Ever been to a conference and took photos of the presenter‘s slides with your smartphone? Or have you ever mapped out a manufacturing process on a whiteboard?
Microsoft has released technology in their Cognitive Services stack called the Computer Vision API, with it, you can supply an image to the endpoint and the API will detect the presence of readable text and transform it into a stream of text! You can see an example of this in the screenshot below:
Being able to point a machine at an image which contains text, rather than input it manually can be a massive productivity boost.
You can read more about this API on the Cognitive Services site here.
Landing.ai is a firm based in Silicon Valley that was founded by AI guru Dr. Andrew Ng. Part of Dr. Ng‘s work at www.landing.ai involves developing machine vision tools to find microscopic level defects in products that simply cannot be identified using human vision. A machine learning algorithm can be trained on a relatively small number of images and yields fantastic results.
The technology has a number of different use cases but is predominantly targeted at the solving challenges in the manufacturing industry which include, but are not limited to:
- controlling and automation
- calibration and tuning
- automated issue identification
Not just content with solving these manufacturing challenges, the startup is also aware that AI-powered technology has the potential to disrupt the manufacturing industry, and with this in mind, the startup is also looking into ways that displaced workers can be retrained.
In this blog post, we‘ve looked at 10 examples of machine vision in manufacturing, we‘ve covered everything from textiles to pharmaceuticals and touched on how artificial intelligence and deep learning are also making an impact into the machine vision space in the form of image recognition.
We hope that by reading this you‘ve got some more insights as to how machine vision can be applied in manufacturing.
Are you using machine vision in your business? Are you considering it?