Needless to say, AI has found its way into various commercial sectors including retail and transportation. Yet increasingly, AI at home has also become a dominant trend. This note examines how AI can help residents secure and optimize their home life.
We’ve been hearing a lot about AI deployed in various commercial sectors. But what about the home?
Before, AI has not received wide attention in the home space
. This however has begun to change, and it’s clear that AI at home will be a major trend going forward. While most smart home products still require some manual programming, already, an increasing number of devices can adjust to user needs based on trends and patterns exhibited by users.
“Although AI is the end goal for smart home brands and service providers, most smart homes still require analog devices like motion sensors and door/window contacts to initiate automations. But some products like thermostats, security cameras and appliance are using AI with minimal human intervention,” said Blake Kozak, Senior Principal Analyst for Smart Home at Omdia
Some AI at home use cases
Security-wise, AI at home use cases include consumer security cameras, some of which now have advanced detection and bandwidth optimization capabilities powered by AI.
“For consumer security cameras, like Google Nest and Arlo, these cameras can identify people, animals, cars and other objects in order to reduce unwanted notifications and false alarms. Soon, cameras will also be able to identify various sounds in the home which could help verify an alarm,” Kozak said. “Also, cameras will be able to operate at lower frame rates and resolution with the help of AI while maintaining a high standard of video analytics. For instance, soon cameras will be able to identify what is most important in an image and focus pixels on those parts of the image, like a face or license plate, effectively lowering the clarity of the surroundings. In other cases, cameras that record 24/7 will be able to use a lower frame rate when streaming to the cloud when nothing important is happening in the frame but could immediately jump to higher resolution and bandwidth requirement if something happens, like a person or vehicle approaches.”
Other examples include home alarms that use AI to reduce false alarms, video doorbells
that can conduct basic conversations with visitors to identify the purpose of the visit and respond automatically; and smart speakers that become more intelligent. “For instance, smart speakers will be able to listen for sounds in the home like footsteps, voices, and other sounds when the home is supposed to be empty. In other cases, the smart speakers will be able to interpret tone of voice, prompting different responses from the speaker,” Kozak said.
Aside from security, non-security AI at home use cases also abound, helping make the home safer, more efficient and more comfortable. “For instance, premium refrigerators from brands like Samsung and Bosch can automatically detect food items like vegetables that are placed in the refrigerator. The appliance can then suggest best placement for optimum shelf-life of the produce within the refrigerator,” Kozak added. “The latest smart irrigation controllers can combine soil type, grass type, slope, wind, seasonality and hyper-local weather data to reduce water waste. Water valves in the home can measure patterns of flow and pipe temperature in order to identify water leaks behind walls.”
Moving toward the edge
Currently, AI at home is typically deployed in the cloud computing environment of either the device manufacturer or a service provider. The cloud processes data generated by smart home devices. Data can range from how often a door sensor is open and closed to the state of the motion sensors.
Yet there is a clear trend that this processing will move more and more to the edge
. “This is especially the case with consumer security cameras where processing is done locally, and video recordings are also stored locally, rather than the cloud. Voice assistants (smart speakers) are also trending in this direction have better natural language processing without the cloud,” Kozak said.
Needless to say, with the home becoming more intelligent by way of AI and advanced analytics such as facial recognition, privacy becomes a concern. And manufacturers must address this concern to win the business and trust of users.
“Overall, facial recognition could lead to privacy concerns and paranoia in certain neighborhoods if the AI starts to send notifications to users of people it does not recognize. Likewise, vehicle recognition could lead to unnecessary vehicle profiling so brands deploying advanced facial recognition will need to proceed carefully,” Kozak said.