CEO of AIoT chip company XMOS, Mark Lippett is a technology leader with 25 years' experience in start-up, scale-up and blue-chip companies.
getty
It's clear perceptions of the home have shifted dramatically. In a previous article, I covered how the pandemic, with remote work and hybrid models becoming more prominent, forced us to rediscover the centrality of these spaces to our lives. Owl Labs data from September, for example, suggested 87% of U.S. professionals believe flexibility in where they work is important, and only 29% wanted to return to the office full time post pandemic. This isn't surprising as the realization sets in that Covid-19 is here to stay in some form.
Under these circumstances, it's necessary to consider how technology can adapt our homes to our emerging needs. At this point in time, consumers have been slowly but surely buying "smart" technology for their homes to enhance things like lighting, entertainment and security. But what's next for this kind of smart technology? How can we make these devices even more intelligent and minimize the necessity for direct user input?
The Smarter Home
The answer, I believe, is in "ambient sensing," which I covered in a previous article. In a nutshell, a home with ambient sensing can automatically adapt all kinds of systems such as lighting, security, entertainment and temperature to the specific preferences of the individuals within the home.
But how is it actually going to work? To answer that question, we need to assess how the intelligence behind the smart home of the future could operate and, critically, interoperate. Currently, there are two architectures we could follow.
Centralized Decision Making
One possibility is to use a centralized decision-making hub that's purpose built to process all data streams across the home.
With this model, every room will be equipped with sensor arrays covering all forms of data collection. The central hub will process the incoming data streams and adapt the home to meet user preferences — all based on intelligent analysis of current conditions and past behavior. When arriving home from work, your security system would be able to recognize you and send image data from the camera to the central hub. If it knows you like a cup of tea when arriving, it could then turn the kettle on by the time you're in the kitchen. A similar process would also play out for your heating, lighting and sound system, for example. This approach provides a highly personalized experience for the consumer.
The good news for device manufacturers is that they won't need to completely redesign their products to take advantage of this new level of smartness. The downside, however, is that it's also more likely to be expensive, given the need to install multiple sensors across the home — and any problems with the central controller mean the entire installation can be put at risk.
There's also some uncharted territory when it comes to identifying who, exactly, will manufacture this central hub. Regardless, with our devices becoming ever more intelligent, it's essential that prospective developers work with device manufacturers to ensure different devices can be integrated into one smart home.
Distributed Intelligence
The second possible architecture for the smart home is "distributed intelligence," where each smart system possesses built-in sensors and the ability to share information with other smart systems across the home.
Information derived from sensors will flow between devices directly, so when the audio sensor on your home entertainment system detects the sound of glass smashing at 3 a.m., information will be communicated to your security system, which will decide whether or not to alert you of a possible burglary or even call the police if your home knows that no one is legitimately present.
A distributed model will streamline installation as the lines of responsibility are clearer than with a centralized hub, as each system is responsible for collecting data and communicating information. However, this architecture will also require interoperability between devices issued by different manufacturers, who will need to agree on a standard for data sharing and processing.
Why Edge AI Works For Smart Homes
I believe both models and hybrids are viable for smart homes going forward, but engineers need to assess which is more efficient before solutions are brought to market. Whatever the outcome, it's undeniable that manufacturers will need to adopt edge AI to make smart homes more responsive. With the incorporation of this technology, all data will be processed locally within the home at a faster rate. This will minimize the need for cloud solutions and the associated hacking risks, latency and scalability concerns.
The incentives to produce functional AI architectures are apparent, with business leaders and industry professionals already acknowledging the potential of this technical breakthrough. The global edge market is already set to reach a valuation of $17.8 billion by 2026, and 77% of the engineers surveyed in my company's 2021 Edge of Today report agree there's a greater need for edge processing to meet the growing needs for information required for our home devices to deliver real value.
With interest in alternatives to the cloud beginning to boom, we can expect a huge rise in demand for flexible and affordable chip technology that will really embed intelligence into the home. Electronics manufacturers should be keen to improve the domestic experience now that we're spending more time indoors, and edge AI is the most viable way to do it.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
