Making your space smart with AI
Recently at AIROUND we’ve had an exciting experience using both IoT and AI in our project for smart city and home automation: it is a personalized microlocation system called HomeKontrol, designed to automate residential and commercial real estate by predicting user actions.
HomeKontrol is a new stage of evolution in the relations between humans and their living space; your living space learns to do it for you by watching what you do and drawing conclusions. Just like remote controls appeared in the past, followed by systems capable of reacting to simple events, now machine learning and micropositioning help a smart home to understand its owners’ preference. It automates routine processes and performs actions before the owner even realizes that they are needed.
Smart or intelligent homes can make rational decisions, introducing automation to various spaces. This is done to improve the experience of residents or employees and reduce operational costs.
Recording and predicting patterns of residents’ location (in the future it will be possible for inanimate objects, too) using microlocation and connecting all devices and switches into a single system makes smart homes not only possible, but also efficient. This requires a certain set of hardware and machine learning algorithms.
Moreover, Episode Discovery system helps establish the frequency and relevance of certain actions within templates and make an informed decision concerning launching a certain chain of events.
A smart home can be defined as a living or working space that naturally interacts with humans and adapts to them.
By adaptation we mean that a smart home learns to recognise behavioural patterns of a specific person and make changes depending on the owner’s current activities with minimal human involvement.
Thus, a smart home unit must be able to predict movements and uses of equipment and devices. By using such predictions, a smart home can display relevant information, play media files, and automate actions that we are used to carrying out ourselves. A smart home develops its own processes by monitoring current conditions using sensors and acting accordingly via a system of effectors.
Let us examine the following scenario.
It’s 7 am on a cold day in January; alarm rings; our homeowner John wakes up and gets out of bed. The light comes on automatically in his bedroom, and the coffee machine is turned on. John heads to the bathroom, while latest news and a weather forecast are displayed on a special screen there. As John finishes brushing his teeth and goes to the kitchen to drink his morning coffee, the light and the screen in the bathroom are switched off, while a screen in the kitchen starts displaying news. As John gets dressed and leaves the apartment, all lights are turned off together with the TV, the door locks itself, and even the iron (which John had accidentally left plugged in) switches off and cools down. While John is away, his smart home places an order with a local supermarket for milk and bread; when John comes home, the order has already been delivered.
HomeKontrol records all the details of John’s interaction with the apartment’s infrastructure to make all this possible. No need to call your virtual assistant.
Everything is a source.
It is possible to predict actions carried out not only by people but also by devices. For example, a remote control with a micropositioning chip can become a universal tool to manage all electronic devices at home, depending on where the remote control is located and at which device it is pointed at.
Virtually any object, even if it does not use electric energy, can become a source of data, trigger, and SmartSpace controller once equipped with a microlocation chip.
A mug, clock, dog collar — they all can feed information into the system, which will analyse the data flow together with human actions and location, followed by a command to carry out various actions. The SmartSpace ecosystem can cover all areas of everyday activity.
It is possible to use additional devices — for instance, temperature and humidity sensors. Data received from them will make the machine learning and forecasting system more intelligent: for instance, it might be possible to turn on bathroom ventilation when humidity exceeds a certain previously studied level, or record a resident’s preferred temperature and adjust the air conditioning system accordingly using positioning.
The system can also prove interesting for businesses from the physical microtargeting point of view: multimedia screens that display highly targeted ads to people located next to the screens; personalised offers at neighbourhood stores; utility companies that can save electric energy based on microlocation data, etc.