A Step Towards Personalized And Automated Smart Homes.

Automated Smart Homes:

The development of automated systems that track occupants and adapt to their preferences is the next important step for the future of smart homes. When you enter a room, for example, a system could adjust to the temperature you prefer. Or when you sit on the couch, a system can instantly move the TV to your favorite channel.

But allowing a domestic system to recognize occupants while moving around the house is a more complex problem. Recently, systems have been constructed that locate human beings by measuring the reflections of wireless signals from their bodies. But these systems can not identify individuals. Other systems can identify people, but only if they always carry their mobile devices. Both systems are also based on tracking signals that could be weak or blocked by several structures.

MIT researchers have built a system that takes a step towards a fully automated smart house by identifying occupants, even when they do not carry mobile devices. The system, called Duet, uses wireless signals reflected to locate people. But it also incorporates algorithms that ping nearby mobile devices to predict the identities of individuals, according to who used the device for the last time and their predicted trajectory of movement. It also uses logic to find out who is who, even in areas without a signal.

"Smart homes are still based on explicit application information or telling Alexa to do something. Ideally, we want homes to be more reactive to what we do, to adapt to us, "says Deepak Vasisht, a PhD student at the Computer Science and Artificial Intelligence Laboratory (CSAIL) and lead author on an article describing the "If you enable recognition of the location and identification recognition for smart homes, you could do it automatically." Your home knows that you walk, and where you walk, and you You can update yourself. "

Experiments conducted in a two-room apartment with four people and an office with nine people for two weeks showed that the system can identify people with 96 percent and 94 percent accuracy, respectively, even when people do not They had their smartphones or they were in blocked areas.

But the system is not just a novelty. Duet could potentially be used to recognize intruders or ensure that visitors do not enter private areas of their home. In addition, says Vasisht, the system could capture information about behavioral analysis for health care applications. Someone suffering from depression, for example, may move more or less, depending on how they feel on a given day. Such information, collected over time, could be valuable for monitoring and treatment.

"In behavioral studies, you care how people move over time and how people behave," says Vasisht. "All these questions can be answered by obtaining information about people's locations and how they are moving."

The researchers anticipate that their system would be used with the explicit consent of anyone who is identified and tracked with Duet. If necessary, they could also develop an application for users to grant or revoke Duet access to their location information at any time, adds Vasisht.

The coauthors of the article are: Dina Katabi, professor of Electrical Engineering and Computer Science Andrew and Erna Viterbi; the former CSAIL researcher Anubhav Jain '16; and CSAIL doctoral students Chen-Yu Hsu and Zachary Kabelac.

Monitoring and identification:

Duet is a wireless sensor installed on a wall about one and a half feet squared. It incorporates a floor map with annotated areas, such as the bedroom, the kitchen, the bed and the sofa in the living room. It also collects identification tags from the occupants' phones.

The system is based on a device-based location system created by Vasisht, Katabi and other researchers that tracks individuals in tens of centimeters, according to the wireless signal reflexes of their devices. It does so by using a central node to calculate the time it takes signals to hit a person's device and travel back. In experiments, the system was able to identify where there were people in a two-room apartment and in a cafeteria.

The system, however, was based on people who carry mobile devices. "But in the building [Duet] we realized that you do not always carry your phone at home," says Vasisht. "Most people leave devices at desks or tables and walk around the house."

The researchers combined their device-based location with a device-less tracking system, called WiTrack, developed by Katabi and other CSAIL researchers, which locates people by measuring the reflections of wireless signals from their bodies.

Duet locates a smartphone and correlates its movement with the individual movement captured by the location without a device. If both are moving in closely correlated trajectories, the system matches the device with the individual and, therefore, knows the identity of the individual.

To ensure that Duet knows someone's identity when they are away from their device, the researchers designed the system to capture the energy profile of the signal received from the phone when it is used. That profile changes, depending on the orientation of the signal, and that change is assigned to the trajectory of an individual to identify it. For example, when a telephone is used and then left, the system will capture the initial energy profile. Then, you will estimate what the power profile would look like if a nearby person were taking you along a road. The closer the profile of the changing energy is to the trajectory of the person in motion, the more likely it is that the person will have the phone.

Logical thinking:

A final problem is that structures such as bathroom tiles, television screens, mirrors and various metal equipment can block signals.

To compensate for this, the researchers incorporated probabilistic algorithms to apply logical reasoning to localization. For this, they designed the system to recognize the limits of entry and exit of specific spaces in the house, such as the doors of each room, the headboard and the side of a sofa. At any time, the system will recognize the most likely identity for each individual in each boundary. Then it is inferred who is who by elimination process.

Suppose that an apartment has two occupants: Alisha and Betsy. Duet sees Alisha and Betsy enter the living room, combining the movement of their smartphone with their trajectories of movement. Both leave their phones at a nearby coffee table to be loaded: Betsy goes to the room to nap; Alisha stays on the couch watching television. Duet deduces that Betsy has entered the limit of the bed and did not come out, so she must be in bed. After a while, Alisha and Betsy move into the kitchen, let's say, and the signal falls. Duet reasons that two people are in the kitchen, but does not know their identities. When Betsy returns to the living room and picks up her phone, the system automatically re-labels the person as Betsy. By process of elimination, the other person still in the kitchen is Alisha.

"There are blind spots in homes where the systems do not work, but because it has a logical framework, you can make these inferences," says Vasisht.

"Duet adopts an intelligent approach to combine the location of different devices and associate them with humans, and takes advantage of localization techniques without devices to locate humans," says Ranveer Chandra, a principal investigator at Microsoft, who did not participate in the work. "Accurately determining the location of all residents in a home has the potential to significantly improve the user experience in the home. ... The home assistant can customize the answers according to who surrounds them; The temperature can be controlled automatically according to personal preferences, which translates into an energy saving. Future robots in the home could be smarter if they knew who was where in the house. The potential is infinite. "

Next, the researchers point to Duet's long-term deployments in more spaces and to provide high-level analytical services for applications such as health monitoring and smart homes with responsiveness.