Home Computer device Q&A: Dina Katabi on a “smart” house with real intelligence | ...

Q&A: Dina Katabi on a “smart” house with real intelligence | MIT News

5
0

Dina Katabi is designing the next generation of smart wireless devices that will sit in the background of a given room, gathering and interpreting data, rather than being wrapped around the wrist or worn elsewhere on the body. In this Q&A, Katabi, le Thuan (1990) and Nicole Pham Professor at MIT, discuss some of his recent work.

Question: Smartwatches and fitness trackers have given us a new level of personalized health information. And after?

A: The next frontier is the home and building of truly smart wireless systems that understand human health and can interact with the environment and other devices. Google Home and Alexa are responsive. You tell them to “wake me up”, but they sound the alarm whether you are in bed or have already left for work. My lab is working on the next generation of wireless sensors and machine learning models that can make more personalized predictions.

We call them the invisible ones. For example, instead of sounding an alarm at a specific time, the sensor can tell if you’ve woken up and started brewing coffee. He knows how to silence the alarm. Likewise, he can monitor an elderly person living alone and alert his caregiver of any change in vital signs or eating habits. More importantly, it can work without people having to wear a device or tell sensors what to do.

Question: How does a smart detection system like this work?

A: We are developing “non-contact” sensors that can track people’s movements, activities and vital signs by analyzing radio signals bouncing off their bodies. Our sensors also communicate with other sensors in the home, allowing them to analyze how people interact with devices in their home. For example, by combining the location data of users in the home with the power signals from household smart meters, we can know when devices are in use and measure their energy consumption. Either way, the machine learning models we are developing in conjunction with the sensors analyze radio waves and power signals to extract high-level information about how people interact with each other and with their own. devices.

Question: What is the hardest part of building “invisible” detection systems?

A: The range of technologies involved. Building “invisibles” requires innovations in sensor hardware, wireless networks, and machine learning. Invisibles also have strict performance and safety requirements.

Question: What are some of the applications?

A: They will make it possible to create truly “smart” homes in which the environment perceives and reacts to human actions. They can interact with home appliances and help homeowners save energy. They can alert a caregiver when they detect changes in a person’s health. They can alert you or alert your doctor when you are not taking your medications correctly. Unlike portable devices, invisible ones don’t need to be worn or charged. They can understand human interactions, and unlike cameras, they can collect enough high-level information without revealing individual faces or what people are wearing. It is much less invasive.

Question: How are you going to build security into physical sensors?

A: In computer science, we have a concept called challenge-response. When you log into a website, you are asked to identify objects in multiple photos to prove that you are a human and not a bot. Here, the invisible understand actions and movements. Thus, you may be asked to make a specific gesture to verify that you are the person being monitored. You may also be prompted to browse a monitored space to verify that you have legitimate access.

Question: What can invisible things measure that wearables can’t?

A: Handheld devices track acceleration but do not understand actual movements; they can’t tell if you went from the kitchen to the bedroom or if you just moved around. They can’t tell if you’re sitting at the dinner table or at your desk at work. The Invisibles tackle all of these issues.

Current deep learning models are also limited, whether the wireless signals are collected from portable or background sensors. Most deal with images, speech and written text. In a project with the MIT-IBM Watson AI Lab, we are developing new models to interpret radio waves, acceleration data and some medical data. We train these models without labeled data, in an unsupervised approach, as non-experts have difficulty labeling radio waves, acceleration, and medical signals.

Question: You have founded several startups, including CodeOn, for faster and more secure networking, and Emerald, a health analysis platform. Any advice for future engineer-entrepreneurs?

A: It is important to understand the market and your customers. Good technology can make great businesses, but it’s not enough. Timing and the ability to deliver a product are essential.


Source link

LEAVE A REPLY

Please enter your comment!
Please enter your name here