Sensors fuel race for enhanced smart-product analytics
SAN JOSE, Calif. — A startup with a brainwave monitor powered by deep learning could help mobile and cloud-computing giants tap into your thoughts. The company was one of several device makers that responded at the Sensors Expo to a call from Google for smart home devices to support its Assistant service.
Cloud services and smartphones probably won’t be monitoring thoughts anytime soon. However, the industry is clearly in a race up the food chain from core sensor technology to software and data, according to several vendors at the event.
Petal uses off-the-shelf headsets that measure electroencephalogram signals that it processes with homegrown neural-network models written in Google’s TensorFlow framework. The two-year-old startup is currently demoing software that lets users play video games with their thoughts.
The company said that Samsung responded positively to a demo that it conducted at Samsung’s Korean headquarters. The startup is seeking connections with Google’s home and embedded groups. The five-person company hopes to ship a developer’s kit within a month supporting Android, iOS, the Mac, and Windows.
For its part, Google wants to expand its ecosystem of 35,000 devices certified to support the APIs for its Assistant service. Assistant is anchored in Google’s HomeGraph, a software construct that mines smart home data for insights.
The service is “based on voice today, but the future is about contextual understanding and visuals, and that’s where more sensors come into play,” said Wendy Kam, a Google business development manager speaking at the event. “More things need to include sensors for the Assistant to offer a robust experience” that lets consumers personalize predictive services.
So far, Google’s Assistant partners include Philips Hue in lighting and LG and iRobot in appliances, but “we are at the early stage of the Assistant ecosystem,” said Kam. “If you want your sensor to work with other devices, let us know the sensor and use cases so we can define traits and APIs.”
Google competes with Apple, Amazon, and others to create a smart home experience. In China, Baidu was early to launch a similar effort. To date, no standards have emerged for devices to support a hybrid mix of services.
Google aims to launch a home security service soon, roughly similar to Amazon’s Alexa Guard. The search giant is also driving a TensorFlow Lite initiative to put deep-learning models running on its own controller in just tens of kilobytes into low-latency embedded services that may not need to be linked to the cloud.
Long term, the kinds of third-party add-on products that consumers deploy on doors and windows for security will become embedded in new homes and furnishings. “Companies come to us asking how to future-proof their businesses, and we recommend them to system integrators,” said Kam.
It’s part of a broad trend of leveraging digital technology to create efficiencies and new revenue streams. “Everyone wants to move into the software-as-a-service model,” said one presenter at the event.
>> Continue reading page two of this article originally published on our sister site, EE Times: "Sensors See Big-Data Opportunity."