How smart should a smart home be before it's worthy of the name? To date, perhaps the term has been too readily applied to homes that are merely high-tech. Automated systems, remote control of appliances from mobile devices, TV and phone over IP—these are all welcome breakthroughs. These technologies are almost synonymous with the smart home and so-called intelligent buildings in general, but there's little or no intelligence to them. For a home to be considered smart, it must in a sense become a robot—a machine capable of, if not true intelligence (and certainly not sentience), sensing data, processing it, drawing conclusions of its own accord, and then acting upon those conclusions.
It's a distinction which Diane Cook of Washington State University's School of Electrical Engineering and Computer Science is acutely aware of. Her research into smart homes goes well beyond presence-detecting light switches, IPTV, and automated garage doors. Cook is interested in homes that to all intents observe their residents and make decisions on their behalf for their own wellbeing. In some cases these decisions are simply for the purposes of convenience: one job less for the homeowner or their family. In other cases these may be decisions that, for a variety of possible reasons, the resident is incapable of making on their own. It's research that raises not only possibilities, but ethical questions and difficulties. Ars spoke to Cook about her work, and about the field of research more generally, to find out what sort of decisions our homes may be making for us in the not-too-distant future.
The intelligence, it is ambient
If you recognize this description of human-aware technology as ambient intelligence you're spot on. It's a term Cook readily applied to her work in a recent article in Science. Ambient computing is a theoretical computing paradigm that could follow once computing is truly pervasive and ubiquitous. Pervasive computing goes beyond the idea of a personal computer in every home, or even in every pocket—it's to do with chips and sensors in everyday objects and the Internet of Things. When computers are everywhere, ambient intelligence is what follows—at least in theory: a soft network of computers and devices (sometimes visible and recognizable, sometimes not) watching, thinking, and communicating for the betterment of humankind.
If ambient intelligence is next, it's logical that the home will be among the first places we experience it. But what will this mean? "The idea is that computer software playing the role of an intelligent agent perceives the state of the physical environment and residents using sensors, reasons about this state using artificial intelligence techniques, and then takes actions to achieve specified goals," writes Cook. Such goals include "maximizing comfort of the residents, minimizing the consumption of resources, and maintaining the health and safety of the home and resources."
Better than human?
Cook has observed that most "smart home" technology to date relies on the users creating their own rules to control their lights and heating systems. But Cook argues that the technology is already there to observe occupant behavior, identify trends, and automate on the occupant's behalf. Doing so removes the need for guesswork and intuition. It can result in smarter, more economic use of resources.
A lone resident in an apartment might be great at turning the heating up when they're too cold, and down when they're too hot. But they might be much less adept at setting the heating to the ideal set point for their desired comfort level in the first instance. A smart heating system would not only identify that set point on behalf of the user, but begin to ease up the heating before they've even noticed a chill in the air—depending on the time of year and external weather conditions, of course.
Is the system smarter than the occupant? Of course not. Is it better able to make decisions that will bring down energy bills? Very possibly, if only because it isn't spending processing cycles on what's for lunch, or how to worm out of tomorrow's meeting. This may be a very basic, simple-to-implement example, but that's not to say it exists. The ability to identify, assess, and then act is not widely found in actual homes, Cook asserts. And, Cook told us, complexity is not a prerequisite. "Simpler is better. Allowing your home to automate control of devices does not have to be complex. However, it can allow the home to time expensive operations (e.g., laundry, water heater) when renewable energy is available or automate control of devices to turn them off when they are not needed for your activities."
House M.D.
But in fact the same principles can be applied to the care of the ill or the infirm. Cook points to research that suggests ambient intelligent homes could help to identify the onset of dementia symptoms (by observing, recording and analyzing occupant mobility patterns) as well as screen young children for autism.
"Our project is going one small step further by recognizing and reporting on occurrence of a more detailed battery of activities and correlating changes in behavior with cognitive and physical health. We are also starting to automate assessment of cognitive health, but this is very much in the research stage." And perhaps more intriguing... "We and other groups are designing reminder-based interventions to help individuals initiate important activities and promote healthy behavior."
Your next refrigerator might kindly remind you to go for that run you'd decided to put off until tomorrow (it won't have to ask, it'll know). It won't deposit your weekly Mars bar unless you do. Hopefully less likely is next-gen social housing that can enforce cold turkey regimens on unwilling heroin addicts.
The watcher in your pocket
But if this sounds like the distant future, Cook points out that the technology to monitor occupants through wearable electronics is fast becoming ubiquitous. In fact there's a good chance you have a highly sophisticated monitoring device on your person at this very moment: your smartphone. "Wearable sensors provide a variety of information," Cook told Ars. "Many include accelerometers (as do phones) which provide constant readings of acceleration in the X, Y, and Z directions. Phones also have gyros which help provide orientation."
Cook even suggests that smartphone cameras and microphones may also be put to use as data-gathers. "There are clever ways to share information between the home and the phone so that activity-aware services can be provided anywhere, for example, to monitor your well-being and to automate interventions." That's how your fridge knew you hadn't been for a run: your iPhone told it (and it's not necessarily a given that apps ask permission before mining your data).
With the mining of data on such a scale comes the prospect of it being put to more nefarious ends (both real and imagined). "Most people have a perceived issue with the security of data, because they want to maintain privacy and are not sure what all can be done with information," Cook said. "Some examples of real issues are insurance companies using gathered information to change your rates, 'cyber cat burglars' using behavioral data to know when to rob your home, and identity thieves wreaking havoc."
The needs of the many
But the hardware is available and the software sufficiently robust to implement smart home technology to beneficial ends today, with one very specific caveat. In our simple scenario of a smarter heating system, we were dealing with a lone resident. Ambient intelligence in the home becomes much more difficult when you introduce more occupants, Cook suggests. "Smart home technologies are pretty robust now in terms of data collection, identification of basic activities, and some automation, for simple conditions (one resident, structured lifestyle). For complex situations with multiple residents, pets, guests, interruptions, this gets much harder and is an ongoing research problem." If you want to live in a smart home any time soon, your best bet may be to live alone.
Science, 2012. DOI: 10.1126/science.1217640 (About DOIs).
Listing image by Photograph by Washington State University
reader comments
35