Earlier this week, two semi-related stories caught my eye and got me thinking about a key smart home component that I find has been lacking for the past few years: Presence detection in the smart home.
First up was the news about a new feature for Amazon Echo Show devices called “Show and Tell”. With it, you can ask Alexa what you’re holding, and using object detection combined with the integrated camera, the Echo device will tell you what the object is. It’s a great addition to the Echo’s set of useful tools and was developed as an accessibility feature for the blind. It’s not a stretch to see Amazon expand this capability towards person recognition, similar to how the newly released Google Nest Hub Max works.
Also, this week is an insightful thesis from The Information on smart home platforms: The old idea of an ecosystem focused on screens in the house is actually the wrong idea. Instead, it’s all about which platform controls the cameras, per this takeaway from the article:
“The key factor in major tech companies’ battle for control of the home is turning out to be cameras, not screens. Each of the major platforms has extended their service with cameras into the home, and is trying to leverage their area of strength into adjacent rings of integrated services.”
That’s a pretty compelling point of view and I completely agree that cameras have risen in prominence when it comes to the smart home. But I don’t think cameras are a silver bullet, and in fact, they bring privacy concerns that other solutions may not, especially when it comes to presence detection.
Yes, if you want hyper-personalized information or actions from your home devices, the smart home needs to understand identity for context. If I want my music playlists or favorite Netflix show to fire up when sit down in a specific room, for example, my home’s devices need to know that it’s me who is in the room.
Of course, a camera can do that through person recognition algorithms. And that camera may not even have to be expensive: The $20 Wyze Cam gained basic person detection earlier this year. I wouldn’t be surprised to see it or other low-cost cameras take the next step and offer person recognition features in the near future.
However, this crosses a privacy boundary inside the smart home that not everyone is willing to accept. Unless and until such recognition features happen locally on a camera device and person recognition data isn’t stored in the cloud – where it could be used or sold for various purposes – cameras aren’t the key solution for a smart home to know who is where in the house.
There are several other technologies currently available, yet underused, for the context of presence. All of them are in the phone you likely carry with you just about everywhere, both in and out of the house. And they’re generally less invasive too.
Using GPS, a phone knows where you are at any given point. That’s why many smart home systems support geofencing: A way to create a virtual fenced area around your home that’s used to trigger some smart home action when a specific individual leaves from or arrives in the specified area. My playlist could be fired up when I arrive home, for example, without any camera or facial recognition used. Geofencing can’t handle fine-ranging location detection, which makes it bad for distinguishing when a person is in a particular room but it works for general neighborhood location.
For more precise location data, there’s a specific address for the Bluetooth radio in my handset. By definition, that radio represents me, so my home can indirectly detect my presence by directly seeing the Bluetooth radio signal. That was part of the Bluetooth beacon initiative a few years back, although that implementation seems targeted more at public venues these days. And Bluetooth has traditionally been more about data transfer or device-to-device communication instead of location.
However, the upcoming Bluetooth 5.1 standards will change that. As I noted back in January, Bluetooth 5.1 adds location accuracy down to the centimeter-level, which is more than adequate for presence detection in the smart home. Combined with basic motion sensors, a smart home could know exactly who or how many people are in what rooms, providing the right context for personalized actions without any cameras.
These examples are more optimal to balance utility and privacy. I don’t need a camera watching me throughout the house capturing exactly what I’m doing, what I’m wearing – or not wearing! – or what I’m interacting with.
Don’t misunderstand me: I see the value that cameras and software algorithms provide to the home. But they’re not always the best solution for the private spaces we live in. A smart home should have just enough information to do what’s needed or requested by the people in that home without a full-on invasion of privacy, but not any more than that.