The Smart Home: How Do We Get There
Smart homes hold the promise of revolutionizing our lives, by simplifying or out-right automating many of our daily tasks. But for the vast majority of us that remains a promise — today we have connected but hardly smart homes. I believe there are four major areas we need to advance upon.
1) single sign-on
A truly smart home will allow you to sign on to different devices in one go, even if they belong to different companies. This is not inconceivable by any means, in fact I would argue it’s inevitable given the benefits of a networked system. Consider what has happened on every single computing platform we have developed, from desktops to cell phones, where different hardware and software vendors have coalesced into products that work (for the most part) seamlessly. The caveat is homes electronics have been around for quite a while and we are not creating things from scratch ie there are legacy devices and entrenched players.
2) auto discovery
The smart home needs to be able to discover new devices, beyond those that exist in the market today.
At one level we need to create a world with widely adopted standards and protocols. If you are working with IoT you will most definitely have come across ZigBee, a wireless language for devices to communicate with each other. It holds a pretty decent chance of becoming the standard given the large number of companies that have adopted into the ZigBee Alliance but it is still missing some very big names. The next couple years will be a battle of trade-offs — standards are better for end users but create more competition / commodization for individual companies.
The second important aspect of discovery is sending software updates over the air, similar to what Tesla does with its cars. That way the hardware in a house is current as new technologies come into play. This is hardly the case today and should be seamless to the point that the home dweller isn’t even aware of it.
3) create own rules
The state of art in home automation today is creating rule based systems ie the users enter a bunch of rules that the house follows. This is hardly ideal because it requires sophisticated users and breaks down easily when conditions change even slightly. What you really want is a self-learning system and the devil here is in the details. If you open the door and turn on the light the smart home will learn that opening the door means automatically turning on lights. But then there are so many edge cases — the lights don’t turn on when there is enough sunlight, the lights might stay on when someone closes the door if there are still people in the house etc. Indeed, there are a few patterns in any home that are quite fixed but most vary a lot based on the person, the time and the context. Creating a state machine that can generate rules and constantly update them is a big AI challenge that no company today has really cracked.
This is the piece that will be least visible to home dwellers themselves but vital to the companies behind the devices. Being able to see real-time patterns will allow enterprise to measure, monitor and modify how their products are being used. Many companies today have been focusing on doing this on security, say how sensors inside the house are behaving can give a good idea of whether there is an intruder. Other companies today have been focusing on energy management, say just even timing the operations of different appliances to incur least load and cost on the electric grid. A truly smart home will have mountains of data coming out that are being analyzed and used to optimize its operations.
The Jetsons was set in 2062 so we still have some ways to get there. Thanks to Michael Mendoza for inspiring this article. As always, appreciate any likes, comments and shares!