From grocery stores to video doorbells

MICHAEL WHEELER, EDITOR-IN-CHIEF MICHAEL.WHEELER@PHOTONICS.COM

Even before COVID-19, retailers were turning to vision for better inventory management, theft prevention, and productivity improvements. The pandemic accelerated this trend, and last year Amazon, which runs the most notable of the cashierless stores, Amazon Go (now part of Amazon Fresh), began licensing the unique set of underlying technologies that allow shoppers to walk into a store, select the products they want, and walk out. All without lines or checkouts.

As we learn in this edition’s cover story, “Embedded Vision and AI Transform Retail Shopping,” the company’s Just Walk Out technology includes ceiling-mounted standard color cameras with custom processing boards, and time-of-flight depth-sensing cameras. Vision data is augmented by weight sensors.

Other retailers are using technology from Grabango, which follows a similar approach — incorporating deep learning, a fusion of various sensors, and embedded vision to deliver a seamless and frictionless shopping experience. For insights into how vision is helping to reshape the shopping experience, see here.

Retail shopping is only one sector in which embedded vision is making its mark. Today, you’ll find embedded vision in intelligent video doorbells and in robots that keep watch over vineyards and industrial buildings. And while there have been notable advancements in sensors and optics, perhaps the most notable are in the processors themselves. New chip architectures are improving processor performance and efficiencies, making what was once a demanding and expensive vision task affordable for consumer applications. Contributing editor Hank Hogan’s “Embedded Vision Is Set for Application on a Massive Scale” begins here.

Rounding out this edition are two articles that cover 3D technology. In “3D Imaging Sees Growth in Multiple Dimensions,” science writer Michael Eisenstein examines time-of-flight, structured light, and stereoscopic cameras in 3D technology (read here). In “3D Matching and Deep Learning Transform Bin Picking,” MVTec Software’s Rebecca König examines the combination of surface-based 3D matching and deep learning (read here).

Finally, I invite you to attend the first Vision Spectra Conference, to be held online July 20-22. We’ve curated an agenda that features some of the biggest names in the industry, covering trends in vision-guided robotics, 3D imaging, deep learning, and AI. The preview is here.

I hope to see you all there!
>