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

Although the term “smart home” seems to have been around for years, the industry itself is still relatively nascent. While our current generations of Google Homes and Amazon Alexas have established the smart home, they also embody its limitations.

The technology behind smart devices promises levels of convenience and interoperability that were unsurpassed a few years ago, but current expectations aren’t being fulfilled. These devices are typically standalone, use unsophisticated data and struggle without a home network.

The concept of a smart home still evokes a futuristic building straight from the movies, providing levels of comfort and autonomy beyond our current technological limitations. The mundanity of its real-world counterparts is due largely to our limited ability to interact with them.

Voicing Concerns

Current smart technologies rely on voice commands, limiting their autonomy. Current models represent progress in terms of voice recognition, but they can still be cumbersome and frustrating to use. Interacting with smart devices isn’t yet intuitive or “human” enough to create the seamless experience we seek.

However, the smart home can’t be unlocked just by improving voice. While audio sensors are a crucial element of intuitive interaction with the smart home, they should form part of a sensor array that can provide devices with more contextual information. To make our homes genuinely smart, devices need to paint a more meaningful picture of the home, making contextualized decisions with superior personalization.

Breaking Boundaries

While every device could pack a unique sensor array, some mediums stand out more than others. Imaging alone can provide vast amounts of information, equipping smart systems with new means of perceiving the home. Acoustic sensors, gas sensing and 3-D mapping would provide the information needed to take smart environments to the next level.

Combining these inputs, our devices would be able to better understand and action individual preferences, sensing who enters and exits a room while changing the light, sounds, temperature and safety features to that person’s profile. It’s not just about comprehending the ambient environment but changing it for the better — with or without direct input.

Obviously, that extends beyond creature comforts. Security is one self-evident example given the introduction of motion sensing, 3-D mapping and individual recognition, making our homes safer than ever before. There’s also potential to save energy, with presence, daylight and temperature sensors able to dim lights or modulate air conditioning on hot summer days.

Sensing Challenges 

So what is holding us back from taking this next logical step in the smart home evolution? Its implementation certainly raises technical and ethical questions.

The first issue is consumer privacy. While homeowners have started to become accustomed to smart speakers in their homes, there are already endless examples of data-hungry organizations observing our everyday interactions with those devices. In recent times, Amazon’s Astro robot has been accused of “full-scale data harvesting,” while Ireland’s Data Privacy Commission has raised concerns over Facebook’s smart glasses.

Smarter and more ambient technology in the future will only force consumers to give over even more data than they are currently sharing. Just how much will consumers be willing to offer? How can brands convince customers that their privacy is of the utmost concern?

The second issue is energy consumption. COP26 has reintroduced the need to make electronics more powerful yet energy-efficient into the national discourse. How can that issue be addressed, especially when one might assume that ambient sensing is going to require a lot of powerful continuous processing?

On The Edge

The key to resolving these issues is embedding AI into the devices themselves — imbuing the edge of networks with the intelligence needed to eliminate reliance solely on cloud networks or corporations to capture and interpret data. I’ve previously written about the artificial intelligence of things (AIoT) and the potential impact that edge technology could have, especially on the smart home.

In one fell swoop, the AIoT can address privacy issues and energy consumption. Allowing devices to process commands without the need for external communication or the necessity of cloud connectivity keeps consumers’ personal information local and secure. It also releases the handbrake in performance terms, thereby encouraging consumers to transition from established brand favorites to new high-performance devices. Where energy consumption is concerned, the AIoT’s local processing means that devices can transition from “always listening” to “always ready.” Sensors will only be in use when needed, ensuring that the demand for electricity isn’t continually high.

However, electronics engineers will be well aware of the challenges involved in embedding AI within electronics. Doing so has traditionally been expensive (because AI chips tend to only address high-end AI demands) and complex (because embedding in AI chips often requires a significant redesign of electronics).

Because of these known challenges, most product manufacturers have focused on connecting their devices to the cloud. Data processing in the cloud has been, to date, the only way to imbue the devices in our homes with “smarts.” While it will continue to be a sensible approach for some purposes, the implementation of a new generation of AI chipsets designed for use “at the edge” is now a viable alternative.

A Smart Future 

When seen through the lens of its potential, it should be no surprise that consumers are becoming increasingly frustrated with smart home technology. Increased speed, precise personalization and more cohesive collaboration between devices are all in demand.

A smart home built on ambient sensing at the edge, taking thousands of different measurements of data into account for each decision, is how we facilitate such improvements. Using the combination of sensors to understand our smart home environment and the AIoT to process and action the results is the most realistic way of realizing the experience we want. Indeed, that reality may be much closer than we think.


Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?