Inventor and technologist Haresh Sippy urged those seeking to solve thorny problems to “tackle the root cause, not the effect.” That thinking is at the heart of several of our features this issue, beginning with the cover story, “Analyzing 3D Data for AI-Powered Robotic Automation.” The root cause, in this case, is the quality of the image data itself. Authors Mingu Kang and Hyejin Kim of ARIS Technology note the growing perception that data gathered from 2D cameras and then subjected to deep learning is a viable path to superior quality assurance. But as we learn from the authors, it’s not that simple; for one, it’s difficult to extract feature points from 2D images, a process heavily reliant on data analytics. 3D reconstruction is also sensitive to the noise from data acquisition, requiring additional image processing such as edge smoothing. A better way is to start with robust image data from the beginning. 3D scanners capture a depth of information far beyond what 2D vision sensors can. And when 3D scanning systems are connected to robotic arms, and the image data is subjected to deep learning, the path is cleared for rendering truly autonomous decisions. Read more. The interplay of robotics and vision is a central theme of another of this issue’s features, “Vision-Guided Systems Bring Precision to Fluid Dispensing,” by Johnathan Titone of Nordson EFD. Numerous applications require extremely accurate and repeatable fluid deposit placement. Critical to this function is the vision system deployed, which is increasingly important as robots and their control software get smarter. For the lowdown on which camera types work best in different environments. We round out our features this month with contributing editor Hank Hogan’s look at how innovations in vision systems and machine learning are simplifying parts inspection for automotive manufacturers (read article), and Timo Eckhard’s “Fixing Image Warping Caused by Heat-Induced Turbulence”. Enjoy the issue!