Forget Fido: Your Data is A Better Home Watchdog
Today’s smart homes are still a far cry from their sci-fi movie counterparts.
They don’t collect the newspaper from the front lawn or automatically clean up a mess. But with the rise of data-collecting IoT devices, that dream is closer to becoming a reality—particularly when it comes to home security.
In-home cameras can now do more than simply record what’s happening thanks to AI and machine-learning smarts that allow us to talk to our TVs or trust our heating system to self-regulate. Now, if you came home from work to find a broken vase on the floor, you can ask the house: “Who was in the kitchen today between 9am and 6pm?” The system will serve you footage to reveal the culprit: The family cat.
It’s a step toward the promise of a truly smart home when smart devices work together to keep us comfortable, entertained—and safe.
Burglar or (More Likely) Your Pet Beagle?
With an alert from a traditional home security system, what’s happening is not immediately clear: Did someone break-in, or is it just an outburst from a rambunctious pet? But modern smart security systems leverage cameras, cloud data storage and artificial intelligence to analyze captured data and to discern whether there’s cause for alarm.
“Traditional cameras are great for recording 24/7, but they’re not good at understanding what they see.”
Alex Teichman is CEO and co-founder of AI security camera company Lighthouse.
Cameras with 3D sensors, similar to those used by self-driving cars, can create 3D models of each room. Those precise views allow the systems’ face-recognition algorithms to accurately identify specific people—and even pets. Smart security systems can then push alerts to homeowners when, say, the system recognizes the dog walker who just arrived for Fido’s afternoon walk, or when it detects an unknown person carrying a large TV-shaped object out the back door.
This AI analysis also allows users to ask the camera’s app for specific information (“What did the kids do while I was out yesterday?”) when they need it, rather than sift through hours of footage.
Safe or Suspicious Activity?
Old-school, basic motion alerts near doors and windows provide some security, but modern systems’ smarts provide greater accuracy and, crucially, the ability to do more than merely observe. They learn.
AI algorithms crunch the data they collect, allowing security systems to differentiate between normal and suspicious comings and goings. It might be normal for a teenager to walk through the front door at 4:15 p.m. after school, but an unexpected entry at 10:30 a.m. might trigger an alarm. Many systems are even smart enough to identify common repetitive motions like ceiling fan rotations or fluttering curtains and suppress alerts, says Andrey Katsman, senior director of software engineering at home security company Canary.
And the use cases for this kind of home surveillance expand beyond protection against burglars. Steve Nichols owns a home healthcare franchise, and his company uses modern surveillance tech to help seniors “age in place” in their own homes. The trick? Sensors track movements (and the lack of movement) inside seniors’ homes and alert caregivers to any unusual activity.
“Let’s say at 9 pm, Mrs. Jones opened her front door,” Nichols says. “Immediately an alarm would come up, because that’s not normal. She shouldn’t be out walking at 9 at night.”
Nichols also uses the system to monitor daily routines. When it comes to elder care, he says family members want peace of mind that elderly parents are being well cared for and making it to scheduled appointments. The sensors allow them to receive these positive updates in addition to safety-related alerts.
The Foundation of a Truly Smart Home? Next-Gen Surveillance
In the case of most home surveillance systems today—and Nichols’ is just one example—a push notification prompts a human to act. But in the future, smart homes could begin to act on their own. Canary’s Katsman says this will likely take place through integration with other smart home apps and appliances.
For example, when a homeowner hosts a party, the security system could detect the number of people who walk into the front door. As the crowd grows, the system could trigger the thermostat to lower the temperature slightly. When the party guests leave, the integrated system could then raise the temperate, saving time and energy costs.
These integrations could even respond to potential dangers in the home. If the camera’s AI algorithm detects a potential fire, for example, the system could first send an urgent alert to the homeowner. If the owner doesn’t respond (or dismiss the alert), the system could escalate the incident to a professional monitoring service. With visual confirming of a fire, the service would direct firefighters to the scene.
“The ultimate goal of advanced detection in smart home security cameras is to provide the user with even more accurate and actionable information without them having to ask for it.”
As Katsman puts it, “The more powerful AI-based integrations can help inform more complex chains of interactions between different smart devices, or leverage additional human assisted services.”
This content is produced by WIRED Brand Lab in collaboration with Western Digital Corporation.