Maker develops a smart home system with self-learning capabilities, using sensors hidden in every room.
Undoubtedly, the rise of the Internet of Things has ushered in a big wave of smart devices. Packed with sensors, these gadgets will ultimately revolutionize the way in which people go about their daily lives — at work, in the car and at home. With respect to the latter, a recent Hackaday Prize entrant by the name of Steven points that there are two basic building blocks for constructing a connected home: a sensor network and an A.I.-based control system. In other words, a smart house can only be “smart” when a complete set of information is sent to the A.I., which in turn, figures out what to do autonomously. Otherwise, it just becomes another thing to point a remote control at.
“Instead of automating anything, all they give you is the power of remote control. These systems also have little to no intelligence, and mostly rely on the users to set up every minuscule detail about how the system should operate. As a result, a tremendous amount of work is added to the user (which completely defeats the purpose of a smart home system), and sealing off regular, non-tech users from joining the fray,” Steven writes.
While there are a number of sensors available on the market today, a vast majority are battery-powered and are far too conspicuous to be adorning the walls. With that in mind, the Maker has developed a new approach with his self-learning Squirco Smart Home System that uses a series of sensors hidden within light switches throughout each room.
Why light switches, you may ask? “They are plugged into the mains, which means they never run out of power. They are present in every room, which means the data set will be complete. They are inconspicuous, because they’re everywhere,” Steven explains.
The Maker has his sights on curating a complete set of data that would be provided to an A.I. unit and used to learn in-home behaviors. This basic set of information includes lighting conditions, temperature and humidity and human presence.
The system itself is based on an ATmega256RFR2 along with a Bluetooth Low Energy MCU. With this, Steven has managed to enable automatic smart bulb discovery and pairing, light use pattern learning, precise climate control with his Nest thermostat, presence learning that allows messages to be sent to a smartphone when someone is detected in or around the home, as well as a ‘vacation mode’ which triggers lights to make it appear as though someone is home while away. Beyond that, he has embedded an iBeacon in each switch, along with the learned usage patterns, to put the most relevant lighting control right at a user’s finger tip.
The electronics are all housed inside a 3D-printed case, while powered via microUSB. In order to simplify the user experience, Steven decided to forgo gesture control and instead leave it to pushing the button. He adds, “It was very important to get the feel of the click just right. The click had to have a sharp, tactile feel, and also have the right amount of travel.”
Over the course of his prototyping process, Steven has modified various components, which he elaborately lists in his Hackaday.io page’s log section. Among those tweaks included moving around its LEDs and PIR, repositioning pin headers, and even toying around with an ATtiny85 to control the relay. Want to learn more about the project? Head over to its official page here.