This website uses cookies primarily for visitor analytics. Certain pages will ask you to fill in contact details to receive additional information. On these pages you have the option of having the site log your details for future visits. Indicating you want the site to remember your details will place a cookie on your device. To view our full cookie policy, please click here. You can also view it at any time by going to our Contact Us page.

Machine vision advances

10 May 2021

Julie Busby looks at new trends and new machine vision technologies. 

Continue reading this article

Register now for free and access every article and to register for the print edition.

Despite the obvious challenges of the past 12 months interest in new machine vision technology remains high, providing industrial end users with increasingly intelligent automated systems. This is impacting all sectors – including food production and packaging – as the drive to automate is taking on a new level in the wake of Covid-19 and realisation that labour-intensive processes can be vulnerable to outside influences and also potentially do not offer the most efficient use of resources.

Some of the more recent advances in machine vision can answer the call to automate the most challenging manufacturing and handling processes. Three technology areas in particular are being used in synergy to create powerful solutions – deep learning, 3D and robotics.

Deep learning
Deep learning is based on neural networks, meaning in the case of machine vision, that the system has some intelligence to determine a good product from a bad product, or intelligent sorting, without the complexity of ground-up algorithm development and design required in the past. Multipix Imaging is seeing success for deep learning in some quite specific tasks, such as sorting and picking good/bad products, where the products naturally have variation. 

In a traditional machine vision solution, the system would have been designed to detect good and bad using a ‘golden template’ approach. However, if the product naturally varies but is still acceptable, the ‘golden template’ does not work as it will reject anything that is not near to being perfect. The power of deep learning comes from having the intelligence to recognise that good products can come with variation, just as the natural world expects. 

Imagine combining this intelligence with a robot and you have very flexible, powerful solutions for automated sorting and picking processes.

3D imaging
Another technology area that has continued to gain momentum is 3D Imaging. A variety of sensor/camera technology is creating more clearly defined 3D data, if you like high resolution data, meaning the accuracy finding the 3D object in space is greatly increased which directly results in much improved instructions to the robot for picking and placing applications. Multipix Imaging has been working with Photoneo Phoxi 3D scanner and customers are creating impressive solutions and not only on the traditional ‘static’ approach; until now no 3D vision system has been able to capture moving dynamic scenes without a trade-off between quality and speed. But that is now changing, with the new Photoneo MotionCam-3D which can capture dynamic scenes accurately and fast with the use of Photoneo’s Parallel Structured Light technology which paralyses the scene to acquire multiple images of structured light in one frame. Thanks to the unique technology, MotionCam-3D is able to withstand shocks and vibrations. This means that dynamic scenes or movements of the camera do not cause any distortion of the final output or broken acquisition. The robust performance of MotionCam-3D ensures a high level of detail even in areas that are generally challenging to scan.

New emerging technology and considerable advances in existing technology mean that machine vision is more powerful than ever at addressing the needs of many processes including optimising production, 3D inspection, volumetrics, Intelligent warehousing and robot bin picking.

Julie Busby is owner at Multipix Imaging.

Contact Details and Archive...

Print this page | E-mail this page