Article Contributors:
Nathan Field (Robotics and Automation Manager) | Ysa Gonzales (Marketing Coordinator)| Rhodri Armour, PhD, P.Eng (Program Manager, Industry 4.0) | Maxwell Bauer (Mechanical Engineering Co-op) | Alastair Trower (Director of Business Development)
Want to Increase Production? Machine Vision may be for you.
We’ll cover what it is, the different types, applications, and whether or not your business can benefit from it.
In this article, we'll cover:
Having trouble keeping up with market demand?
Maybe you have a process that’s taking too much time and labour. Maybe you’re not utilizing your labour resources as efficiently as you think you can, and you’re looking to increase productivity.
Whatever the reason, you’re likely considering implementing a Machine Vision system to help you automate your process. But what exactly is Machine Vision and how does it work? What types of problems can it solve? Is it applicable to your business?
And most of all, is it the right business move to make?
What is Machine Vision?
What’s the difference between Machine Vision and Computer Vision?
Historically, vision-based analysis has been a vital tool in industrial applications, leveraging the large amount of visual data present to perform quality control, defect detection and removal, and many other tasks. Traditionally, this has been a role performed by long-serving, experienced staff.
Visual Quality inspection is a role that is now difficult to hire for, as people are less content to be employed to repetitively look at product on a line all day. Thankfully, computer vision systems have advanced sufficiently to relieve much of this pressure.
The term “Computer Vision” refers to the process by which a computer receives, processes, and analyzes a video or picture. Building from computer vision is “Machine Vision”, which is where a computer vision system interfaces with a machine to perform actions based on visual inputs. These actions vary in complexity from activating a basic rejection system to directing an industrial robotic manipulator.
Machine Vision is usually applied in an industrial or manufacturing setting and automates the inspection and analysis of objects using cameras. Simply put, it is a machine’s ability to “see”. [1]
Who can Benefit from Machine Vision?
Machine Vision has been proven to be beneficial in any industrial facility that utilizes a repetitive process.
It is used in a wide range of industries, including:
- Food Production
- Food Processing and Packaging
- Automotive
- Pharmaceuticals
- Rubber & Plastics
- Electronics and Semiconductors
- Manufacturing
If you’re operating within these industries, chances are, your competitors are likely already using some form of Machine Vision System.
In an industrial setting, the most common types of Machine Vision applications include:
- Quality Assurance and Inspection
- Measurement
- Product Identification
- Predictive Maintenance
- Sorting
- Machine Monitoring
- Testing, Calibration and Verification
- Code Reading Applications
- Real-Time Process Control
If your business currently does any of these processes manually, then you could benefit greatly from investigating how Machine Vision can up your game. Your operation, staff and clients will thank you for it.
How does a Machine Vision System work?
And how is it so versatile?
Machine Vision technologies are often made up of the following:
- Cameras and Lenses
- Lighting
- Vision processing system
- Additional sensors
- Communications and controls
How does it work? In a typical assembly line, the camera is often triggered by sensors to collect and process an image. The vision processing system may be programmed to determine whether an object is present or not, verify the location of the object relative to other objects, and then determine its colour, size, shape and orientation. Because of the advanced technology in vision processing systems, they can even read & verify printed characters and decode a standard barcode or QR code.
After the scan, the system then determines what to do with the object. If the scanned object passes the predetermined decision thresholds, then it could be selected to move forward with the subsequent process, otherwise, if it fails, it could be rejected to a different line or container – or whatever preprogrammed action is designed into the solution.
What makes Machine Vision adaptable and versatile is that it doesn’t matter what type of object is being scanned, what size, shape, or color it may be. Furthermore, the respective vision systems are customised depending on your specific application.
What are the Benefits of Implementing Machine Vision?
What’s in it for you?
If you decide to investigate the potential for Machine Vision, here’s what you can expect:
1. Increased Production Throughput
Computer Vision Systems can process data at a much higher rate than a person, allowing for increased production output.
2. Better Quality Control
Computer Vision Systems do not get distracted, tired, or lose focus – resulting in less mistakes over time. These vision systems can also see features imperceivable to the human eye either in scale or those that are outside of the visible light wavelengths. Additionally, because of the automated set up, Machine Vision allows for every object to be inspected instead of a random sample, which drives greater traceability and reduces risks to your clients.
3. Data Permanency and Analysis
Vision data can be saved for further reference, allowing for intelligence to be gained from future analysis. Machine Vision provides data logging, inventory control, and image capture and storage. Data collected now can later be interrogated by Artificial Intelligence “AI” potentially driving additional business improvement.
4. Process Improvement and More efficient use of Labour
Machine Vision Systems, inspection dates, and processing results can be logged and monitored to better identify emerging trends and defects. Furthermore, when it comes to labour, vision systems paired with robotics allows for the automation of tedious roles such as product sorting, inspection, packing, and rejecting. As a result, manual labor can be utilized more efficiently by shifting it to other higher value and more rewarding tasks.
What Problems does Machine Vision solve? (According to Application Type)
1. Single Camera Application
For detecting product and basic quality issues (Typically $20K-$60K)
This type of application is great if you have a simple problem that needs solving. These systems often require only one type of inspection, minimal programming, and basic mechanical interfacing. The elements that drive cost in a project like this are commonly the interfacing of the Machine Vision System with other equipment and the design of extra systems needed for features such as rejection or alarms.
This type of system is especially useful for purposes such as:
- Detecting quality issues
- Mechanical Failures (seal, finish, edges)
- Detection of Substances (Glue, Marking)
- Detecting of product
- Is the product present?
- Has a part been installed properly?
A simple system like this can often be tested in-house and can be mostly, if not completely, tested prior to on-site installation.
Has Enginuity done this type of system before?
Yes, we have. Here are some applications that we’ve solved by implementing a single camera system:
- Inspection of a food product can: to ensure lid has been sealed properly.
This system identifies defects on the quality of sealing and diverts probable problems for manual inspection by a trained eye. - System to check the size and quality of riveting during a mechanical construction process.
In this application, when a rivet is considered ‘out of tolerance’, the system alarms and stops indicating to a technician to check tolerance and adjust machine settings.
2. Single / Multi Camera Applications and Connectivity
For product grading and information gathering (Typically $40K-$100K)
Sometimes the problem at hand is not very easy to solve, or perhaps you want more capability than simply a light that turns on when product is bad or a diverter that separates defects from the main product flow. Perhaps the problem cannot be solved by only looking at one angle, maybe you need more cameras.
As a result, this system is harder to design since there are more variables. Multi-camera systems require more upfront investigation (and cost) but can often deliver more value than multiple discrete systems. Elements that drive project cost would mainly include integration time on site and extra hardware required to complete the task.
Often customers would like to log production data such as production date, quality, product inclusion and pictures taken during production: this allows for traceability and document control. Leveraging technology like Optical Character Recognition (OCR) and Quick Response (QR)/Barcode reading, information can be gathered from these systems as well. When technology like this is added, products can be added to databases and tracked throughout the process.
This type of system is especially useful for purposes such as:
Grading
- Grading products based on known elements (colour, size, shape)
Information gathering
- Reading QR/Barcodes
- Inspecting and Reading text (date-codes, timestamps, lot code, print quality etc.)
- Complex analysis
- Multiple inspections being completed on one image
- Many forms of data being pulled from an image
- Database Connectivity
- Information can be formatted and sent directly to databases (SQL) and in other common formats such as .CSV and .XML.
- Images can be stored in local or cloud-based storage and referred to by production management systems.
Has Enginuity done this type of system before?
Yes, we have. Here are some applications that we’ve done before where we built a Single / Multi Camera System:
- Inspection of multiple failure modes (5-10) on one image
- Image files and SQL data sent to local database with ability to be cloud accessible.
- Size and quality judgement of breadsticks being captured in 12 places in fractions of a second to allow for controlled rejection of failed pieces prior to final packaging.
3. Complete System Design, Robotic Integration, Artificial Intelligence:
For complex systems and applications ($60K-$120K)
Often a vision system is a critical sub-section of one larger system. This is commonly the case in industrial applications; combining a vision system into the production management system drives the cost down significantly and allows for greater Return on Investment (ROI) as part of a larger capital plan.
We regularly find that customers want to complete an automated task, such as “take this product from here to here”, which is easy if the product is not moving and is always in the same position. In more advanced systems which require complete system design, the product could be put onto a conveyor belt to go into a robot cell, then the robot could use a camera to locate the object, pick it up and orient it correctly for packaging.
We have also come up against challenges in which a customer may want a vision system with more intelligence. This scenario may arise when we are trying to determine if what we are seeing is product X, Y, or Z. Using our in-house skillset, we have been able to develop Artificial Intelligence (AI) models to detect and classify images based on previous training data.
Some examples of this type of system include:
- Picking randomly placed products from a conveyor belt
- Sorting product based on metrics
- Once the first metric is determined (colour, shape, size) another could be measured before sorting (for example weighing inside the robot tool).
- Using AI, a product can be identified by classification (for example, determining whether it is good or bad)
- A system of many robots that assemble a product, using vision systems for quality control checks during this process
Has Enginuity done this type of system before?
Yes, we have. Here are some applications that we’ve done before where we did a Complete System Design, with Robotic Integration and Artificial Intelligence:
- A system to quickly sort seafood based on type
- A robot system that assesses the object passing before it, identifies which products are oriented incorrectly, picks them up and orients them correctly for the next step in the process.
Removing unwanted objects from groups of products moving under a camera on a conveyor belt (in a washdown and saltwater environment) - Sorting of dried Sea Cucumbers into different grades.
In this case we reduced the amount of manual lifting for line workers while simultaneously grading and detecting for foreign objects.
Summary
The potential of Machine Vision Systems is far reaching depending on the technology deployed and the level of complexity of the challenge. Each problem is different and will require a different solution.
Here at Enginuity, we have many demonstrator vision systems in-house and can quickly develop a concept. We can generally test a theory before you invest in expensive hardware, and we regard this early-stage de-risking as critical to success of our client projects.
Whether you are looking to automate a manual task, improve product quality, or increase throughput – a vision-based solution, similar or different to those listed above, is likely to be the right fit.
Whatever the case, Enginuity can help you along the way, from challenge identification right through to the delivered/implemented solution.
Happy Automating!
About Enginuity’s Automation Services
Not sure who to call? Enginuity offers a full-range of automation solutions and continuous improvement services.
We’ve automated sea cucumber grading using Machine Vision. We’ve sorted pharmaceuticals too and eliminated manual packing for a fish cannery facility.
We work with small and medium enterprises as well as large corporations across a wide range of industries including Aerospace and Defense, Industrial and Marine, Manufacturing, Product Development, Medical Devices, Food and Beverage, and more. Our cross-functional team has a wide range of expertise, ensuring that we get to the root cause of your problem while evaluating your long-term needs, so that you get the best service possible.
Is Enginuity the right partner for you? You can read about some of our past projects (and engineering case studies) and read why clients love working with us.
Our principles-based, agile engineering over the past 17 years is built upon our core mission of making a lasting impact to our clients and ensuring that they get what they truly need moving forward – no cookie-cutter engineering. No meaningless micromanagement. Just lean, fast, creative engineering leadership that understands and moves you forward to your goals, at your pace.
If you think we’ll be a good fit, we can start that conversation here.
References:
- Control Engineering. 2021. Control Engineering | What is machine vision, and how can it help?. [online] Available at: <https://www.controleng.com/articles/what-is-machine-vision-and-how-can-it-help/> [Accessed 28 October 2021].
- What is machine vision (computer vision)? – Definition from WhatIs.com. (2021). Retrieved 28 October 2021, from https://searchenterpriseai.techtarget.com/definition/machine-vision-computer-vision
- Machine Vision – What you need to know (Overview) – viso.ai. (2021). Retrieved 28 October 2021, from https://viso.ai/computer-vision/machine-vision/
- Introduction to Machine Vision Part 3: Key Parts of a Vision System. (2021). Retrieved 28 October 2021, from https://www.microscan.com/en-us/resources/videos/introduction-to-machine-vision-part-3
Considering to implement a Machine Vision System?
You don’t have to figure it out all by yourself. Looking for solutions? Talk to us.
"*" indicates required fields
