Blog

Artificial Intelligence – JM Canty (Pt.1)

Neal Systems’ general manager, Shane Filer, had a chance to interview Tod Canty Sr. of JM Canty. His company is impacting the industrial world with Artificial Intelligence. 

JM Canty is a leading supplier of process tank visual solutions, supplying everything from sight glasses to LED lighting to particle size analyzers to smart visual camera solutions.  Shane was able to chat with Tod Canty Sr., the current president of the company, as to how artificial intelligence found a home with the Canty visual cameras. 

INTERVIEW:   

Shane: Tod, what initially made you realize that artificial intelligence could be applied with your camera systems?

Tod: I remember it was 1988.  I was involved with a project at Dow Chemical, and my contact there wondered what could be done to replicate an operator physically looking into the tank 24 hours a day.  We already had an analog camera that could digitize into pixels, and I knew that machine vision technology was already starting to be used in certain manufacturing sites.  It made me wonder whether the data from the camera pixels would tell me anything. I used BASIC and assembly languages to compile data algorithms, and realized that we could actually “see” differences in data that we could use to mimic an operator’s decision making based upon things like level, color change, and other things happening in the tank.  Thus the Canty Vector Control project was born. 

Shane: What were some of the initial ways that you utilized that artificial intelligence with your Vector Control solution?

Tod: Two that stood out were both Foaming control and Level control.  We realized that we needed better lighting in the tanks in order to be able to better “see” what was happening, and differentiate pixels from light to dark.  We were able to teach the vector control device to digitize the data to recognize patterns as they changed to recognize pixel differences that corresponded to definite level and foaming effects. The Vector Control’s Artificial intelligence had to recognize not just pixel change, but also the rate of change, as well as contrasts with edges (light and shadow).  We kept teaching the algorithms, but realized that we had to control the lighting AND the camera AND the software – all were critical. 

Shane: What are some of the creative ways that you have seen customers teach the system that Canty hadn’t thought of?

Tod: Seeing things with clear water was easy, but agitated tanks were pretty difficult and required more algorithm training, as well as decision trees for complex problems.  We also had people start to use the cameras for particle sizing, high temperature cameras for burners and pilots for thermal oxidizers, as well as looking at the levels of molten glass – things we never considered.  If an operator can see something visually and know a process well enough to make decisions, we can probably use the software to teach it to make the same decisions.   

Shane: Last question – any specific industries that utilized the Vector Security hardware / software to achieve process optimization such that it fundamentally changed the way that they operated the process?

Tod: One of the exciting ones is that is especially prominent right now with the whole push for the CoronaVirus Vaccine is bioreactor batches.  Foaming is an issue that affects just about every industry with their agitated tanks.  Some of our customers that had bioreactors utilized the Canty Foaming Camera, and reported back to us that one very unexpected positive consequence came out of it.  Not only did they have a much faster response to foam as it started (meaning that less foam formed, and less anti-foam chemical had to be added and subsequently removed), but they found that they could now produce 30-50% more product per batch.  They no longer had to leave so much room for the expected amount of foam that they often got, and as such were able to put more product into the tank.  So they installed the camera to save on anti-foam (which they did), but the real benefit became something unexpected – the significant extra amount per batch that could be made. 

Shane: Thanks so much, Tod – greatly appreciate you taking the time to walk through how artificial intelligence actually got programmed in, and excited to see what new ways that industries will utilize the Canty visual solutions!

More information on the JM Canty process cameras with artificial intelligence can be found here! If you would like to discuss your specific tank application and how these process cameras can benefit your process, please reach out to Shane at sfiler@nealsystems.com.