Every other year the Machine Vision industry looks forward to one of the most important trade shows for us in the industry – The Vision Show. This year, The Vision Show 2018 seems to be a little bigger than in the past, with 150+ exhibitors in Boston.
As with most industry and tradeshows, The Vision Show 2018 highlighted trends in the industry with conferences and speakers. The latest trends in the industry seems to be in Embedded Vision, Industrial Internet of Things (IIoT), Collaborative Robots (Cobots), to just name a few.
Instead of focusing on the usual highlights of the show, we sought to write about our own top 3 takeaways and industry comments for this year:
1 – Embedded Vision
The show is packed with vendors for the usual items that you’d expect for a Machine Vision show – cameras, IO boards, lenses and optics, lighting and exotic illumination sources. One underlying theme and buzz-word that was highly visible is “Embedded Vision”.
What exactly is Embedded Vision? According to Keox and simply defined, it’s anything that can do machine vision with a small form factor. The reason we like keeping this definition so simple is that Embedded Vision can be done with anything from small palm-size PCs (eg. x86 based) to Raspberry PI to FPGAs (Field Programmable Gate Arrays) or even ASIC chips (Application Specific Integrated Circuit). Additionally, it should be able to acquire images (ie. camera with lens) which can be anything as small as a board level sensor. Integrating it all together could be everything onboard one PCB to a small case with the components mounted inside. Of course there are other definitions to include on what is Embedded Vision, but that’ll require much more space to discuss.
Machine Vision traditionally required major computing power along with large camera components. With electronics becoming smaller and more powerful, it is starting to handle vision applications in those small form factors – think of the processing power on your mobile phone. As a matter of fact, small embedded vision is a requirement for other developing technologies such as drones and autonomous vehicles, to name a few.
Although “Embedded Vision” is a buzzword that gets everyone’s attention, there are some manufacturers that have components that are more suitable to what is in our mind for Embedded Vision. Just focusing on cameras, board level cameras such as Basler’s dart Board-Level Cameras (www.baslerweb.com) is one popular component to integrate with. Ximea (www.ximea.com) offers great board level cameras along with some crazy small subminiature component cameras (15 x 15 x 8mm in size)!
2 – Software
Machine Vision Systems certainly need optical components like cameras and lenses and other supporting hardware. One element that is absolutely required but not talked about as often is software, and in the context of Machine Vision, image processing libraries. This is certainly not a “one-size-fits-all” choice. Since Keox Technologies develop applications mainly in C# and have our own proven application framework libraries, we look for software that can integrate well with our programming workflow.
Although we employ open source libraries and even write our own image processing tools, we utilize MVTec’s Halcon (www.mvtec.com) for more complicated image processing steps. Speaking of Embedded Vision, they also have Halcon Embedded which can download to various embedded operating systems such as Linux, VxWorks, QNX, Windows Embedded Compact, etc.
A new find for the industry is Innomiles’ OVK Framework Modules (www.innomiles.com), which is based in Taiwan. Their machine vision software modules contains useful tools such as Distortion Calibration and Flat Field Correction tools, along with other valuable tools such as thresholding, blob analysis, and even OCR and Pattern Matching tools – it’s more sophisticated than it first lets on!
3 – 3D Vision
Now that Embedded Vision is a trending topic and that there are software to control it, what’s a good application that would need Embedded Vision? All, 3D Vision sensors on the market can be considered as a form of “Embedded Vision”. They are all somewhat self-contained units that have image sensors and usually contain onboard processing in order to report out the 3D data. 3D Vision is finding more uses in both the manufacturing industry (robot guidance, bin picking and sorting, etc.) and “non-factory” applications such as drones and autonomous vehicles.
There are several measurement techniques to do 3D vision – laser triangulation, structured lighting, and time of flight are some of them – and for each technology there are dozens more companies offering their sensors. What Keox Technologies look for is good, smart, and rigorous engineering of 3D sensors no matter the measurement technique. Photoneo’s PhoXi 3D Scanners (www.photoneo.com) offer a well-designed scanning sensor which also happens to be sleek and visually appealing. Another company with great engineering research is Ajile (www.ajile.ca) and their DepthScan 3D Imaging System. Utilizing their own DMD Projector to generate structured lighting, their system is engineered to be flexible enough to achieve optimal quality tailored specifically for the application requirements.