Home Together, a CES demo

by Pierre Salaün | Jan 06, 2016 | Use Case | IoT  Learning  Smart Home 

Home Together, a CES demo

Meet Gisele, she lives in a nice house, but it’s not just a house. Gisele’s place is empowered with AI. Thanks to a handful of connected devices and craft ai agents ruling over them, her house reacts to and learns from its occupants. A truly smart home.

Context-aware automation

Let’s start with basic automation. For instance, depending on the outside light intensity, the house will adjust the light of the room in which Gisele is. She doesn’t have to turn any switch on or off: she is localized inside the house by an indoor positioning system, such as a Beacon, and the connected light bulbs, such as a LIFX or Philips Hue, react accordingly, following simple logic defined in craft ai.

In our scenario, Gisele likes to have the lights off during the day. When it’s getting dark, she likes to have a low intensity white light wherever she goes. At night, she prefers a bright, warm light.

She does not have to switch any light, the house knows what to do.

Preferences learning

But Gisele does not live alone in this house, you live in there too! And thanks to the craft ai powered home, everybody can have it their way with the connected devices.

As you enter a room in the house it will react in the same way it does for Gisele, her long time occupant. But you probably have different taste. As you adjust the lighting using the regular switches and knobs, the craft ai agents running the house will notice and remember the change and its context. The next time the same context occurs: you entering the room while is dark outside, the light will be set to the setting you previously used. Thanks to craft ai the house learns about your preferences.

When the two of you are in the same room, you can choose to have another setting that is neither yours nor hers. At least, this is the logic that has been defined in our scenario, but we could imagine other possibilities to manage multiple presences in a single room, such as keeping the light settings to those of the first person in the room, or switching the color to one that is in between the many people’s color of choice.

Leveraging more devices

All of the above simply illustrates what happens with the lights, the position of the occupants and the outside light intensity, but we can go a level higher by adding more objects!

If additional connected devices are available, the craft ai agents can take them into account. For example, when the TV is on, it will turn off the light in the living room and dim the light in the adjacent rooms whenever it should be lit.

If we want Gisele to reduce here water consumption, you can use a Hydrao connected showerhead or simply measure how much time she spend in the bathroom. If she’s too long, the light bulbs will switch to an ‘alert’ mode, flashing blue and red.

Thanks to craft ai architecture it is pretty easy to add these additional rules to the lighting controller by composing simple basic behaviors, our building blocks for agents.

We are showing this house at CES!

This scenario showcases three challenges a developers faces when creating an actually smart home:

  1. Implementing simple decision-making that’ll fit basic needs without the need for lengthy setups;
  2. Being able to learn the users preferences by just letting them interact with their devices;
  3. Making it easy to mix and match the particular type of devices the user have connected.

craft ai is particularly well suited to tackle these:

Last but not least our integration with IoT platforms such as La Poste “Hub Numérique” (in french) makes connecting to the different devices truly straightforward.

This is not just a fantasy scenario, we are actually showcasing this demo at CES this week, on La Poste booth, Tech West, Sands Expo, Level 2, Booth 71125. We would be happy to let you visit the house!

If you happen to not be in Las Vegas, we will also shortly release this demo and its sources online and publish a follow-up article to explain more technical details, stay tuned!

Read the follow-up post

Share