An article writen by Paula Doe for SEMI | MSIG and the MEMS & Sensors Executive Congress, 2017 edition - New sensor technologies, and smarter ways of integrating more intelligence, continue to generate unexpected opportunities in the changing MEMS business.

Changes needed for sensors for context awareness

If digital assistants are ever going to be really useful, they’ll need some particular sensor capabilities to understand emotion, suggests Lama Nachman, head of Intel’s Anticipatory Computing Lab, who will give the keynote at the SEMI | MSIG MEMS & Sensors Executive Congress. “What we’re trying to do is change the way we assist people by having a deeper understanding of their needs. So far the assistants that try to be proactive don’t work very well, because they don’t understand the context well enough.”

One problem is that the existing sensors designed for other applications don’t quite measure the right things for these applications. Nachman notes the problems of trying to detect pupil dilation, which correlates with alertness and attention, with the typical camera or image sensor that’s designed for taking a nice picture, or of trying to identify what’s happening in the background with a phone that wants to get rid of the background noise. “It would be very helpful if there were APIs for sensors so developers could set them to do particular things,” says Nachman. “With cycles of innovation moving so fast and machine learning emerging, sensors are not changing fast enough to keep up.”

The real game changer for context awareness will be the ability to recognize emotion, such as with a sensor that measure heart rate variability very accurately, in a form factor that people would actually be willing to wear, she notes. “Recognizing emotion is even more interesting for systems that learn, since emotional reaction is fantastic feedback for an assistant to learn to be useful and not make you want to throw it out the window.”... Full article