Issue link: https://iconnect007.uberflip.com/i/895789
86 SMT Magazine • November 2017 think or as standard as people think. It never is. So, those are the two challenges. Social im- plications are challenging, of course, because if we want to build a plant without people, well, who's going to give you land to build or tax breaks that everybody's asking for these days? It's not going to be from the county or city be- cause you're not hiring anybody. But, some companies can afford to build plants without that, so that will also be part of the decision making. Las Marias: How should these issues be ad- dressed? Vujosevic: I think we have everything right now but machine designs. I mean, we have control- lers, we have software, and machine to machine communication will happen. We have robots, we have automated guided vehicles (AGVs) that we can control for machine delivery, we have sensors, but the most important problem is that machines need to be re-designed. Because noth- ing new has happened in this industry for de- cades. Although, you hear that we lose jobs be- cause of automation. That's not true in the elec- tronics industry. There's no automation at all. Everything is the same as it was 20 years ago. Now, people are introducing robots, but not in many companies. It's going to happen, but there are some technologies, like cloud comput- ing, that's not going to happen any time soon because of security problems. In the U.S., no- body wants to do anything with that in the man- ufacturing industry. Okay, you can share iTunes and Amazon stuff, but nobody in the manufac- turing industry wants to cloud anything. There are some technologies that are well-developed, but the machine designs are the most critical. They need to re-design the machines to support operator-free operation. Las Marias: What about the inspection side? Vujosevic: There are a lot of efforts to provide feedback and feed forward from inspection. Koh-Young is doing some really good stuff about that, for example. Our company now is involved in developing an AI-based system for intelligent process control. Right now, ev- ery test machine has some kind of statistical process control and charting, but nobody uses that. It is left to operators to fix the problem and often the easiest fix is to wash the board and do it again without addressing the under- lying problem. You've got an SPI machine and you see the process chart and no- body even looks at that. When there is a problem with the board, they will bring it back to the screen printer and redo it without ever considering what is wrong with the process. What we need an in- telligent process control where the system will monitor every test ma- chine, every line, and decide and even predict when a process will get out of control, or predict a trend, you know, predict the next measurement, next point, and is- sue alarms or even stop the line. We need that for lights-out electronics assembly, and that can be achieved by an intelligent software. We want to take the operator out of the pro- cess control loop and achieve defect free self-correcting lines. OPTIMAL ELECTRONICS SETS SIGHT ON GROWTH Figure 3: Real-time machine performance monitoring and component traceability data collection.