Home Security – February 14, 2018

How AI will continue to make your Smart Home even smarter

AI, machine learning, and the Internet of Things are all terms getting thrown around a lot these days. But what do they actually mean, and how can they create (or, when misapplied, hinder) the “smart home?”


Deloitte defines AI as “getting computers to do tasks that would normally require human intelligence,” while Stanford University defines machine learning as “the science of getting computers to act without being explicitly programmed.” Whichever way you look at it, machine learning is often a tool for developing artificial intelligence technologies – when computers learn, they’re both exhibiting and developing traits usually reserved for human intelligence. The Internet of Things is a broad term for the massive number of devices that are connected or can connect to each other and the internet, a group that includes everything from FitBits and smart speakers to automated pet food dispensers.

The natural synchrony between these ideas is apparent. A “smart” home that integrates AI and machine learning with networked devices can learn its residents’ habits and needs and use them to shape how the devices meet those needs. IOT, AI, and machine learning technology can combine to create smart thermostats that can anticipate how you’ll want to adjust your home’s temperature throughout the day or speakers that can pick your favorite albums before you even think to play them.

A truly smart home recognizes not only that everyone has different needs, but also that everyone’s needs are constantly changing. Networked AI devices are the way for smart homes to address human fluidity.

But a lot of “smart” home devices aren’t actually that smart. They can connect to the internet to store and send data, sure, but the benefits they provide have to be weighed against the disadvantages of having to update how the home uses them as user preference shifts. The constant struggle to keep “smart” networked devices working in accordance with how people actually live requires an endless stream of adjustments and manual updates. Anyone trying to keep up could conclude that it’s better to be a luddite.

Truly smart homes need to integrate machine learning – they need to reflect both our current habits and how our habits change. Angee is more than just a home security device; she’s also a student. The more time she spends in your household, the more she learns about how your household functions and the better she can protect your home.

The Angee platform involves one central Angee with a moveable camera and any number of security tags that can be placed at entrances and access points, such as windows. When an authorized user – someone you live with – enters or exits the home, Angee takes note of who’s present. When different sets of people are home, or when no one is home, Angee automatically switches into an appropriate security mode.

When no one is home, Angee is on high alert and will notify remote users if any non-authorized people are detected in the house. If family members have returned from work or school, however, Angee recognizes trusted household members and won’t sound the alarm. Angee uses machine learning to pick up on your household habits.

If the house is generally empty by nine in the morning when your spouse leaves for work and picks back up again at four when your kids get back from school, Angee will recognize this pattern and enter higher-alert modes if it’s broken. And if your household pattern changes – maybe your spouse gets a new job, and they need to be out of the house by eight, or school lets out for the summer, and your kids are in and out throughout the day – Angee can learn the new pattern, too. Angee’s machine learning capabilities make her better attuned to when activity in your home is normal and when it’s not, reducing false alarms while keeping your home genuinely protected.

Angee’s audio detection also integrates machine learning components. The Angee camera and security tags contain sensitive audio sensors powered by Audio Analytic’s ai3 sound recognition software. Angee uses her audio sensors to alert users if she detects any suspicious noise, such as glass breaking or a fire alarm.

But Audio Analytic’s ai3 technology is adaptive, meaning it can adjust to the different sound environments it’s placed in. Different homes have different soundscapes, and no one wants a home security alarm that’s constantly getting set off by a normal-for-them noise, such as horns honking or the neighbor’s dog barking. Audio Analytic’s software means your Angee system will know what’s normal or abnormal for your home.

A lot of things could cause changes to your typical day-to-day soundscape – think of what happens when a home gets a new parrot or a new baby! Angee will adapt to this new soundscape too, so you’re always protected while false alarms are minimized. In today’s consumer marketplace, “smart” home devices are everywhere. But for true home protection, you’ll want to invest in a truly smart device like Angee.