tufts universitytufts magazine issue homepage
contact us back issues related links
featuresthink tankplanet tufts Foul Ball Advance Guard Dog Star Clued In to the Brain Brilliant! Line in the Sand D.P.M. Paper Birch Eureka Laurels newswire the big day departments


Robots Who Understand People

“I want to make robots that will be genuine helpers to people. They might be used in eldercare situations, for example—they could make sure older people take their medicine, help them get dressed, get out of bed, do all of the kinds of things a human helper would do. To achieve this, though, we’ll need to able to interact with the robots in the same way we’d interact with another human. So how do we get there?

Well, the first ingredient is natural-language understanding. It’s not enough to just understand words, though—they’ll also have to understand subtle things like gestures, tone, context, and facial expressions. That means tying together lots of different areas of robotic sensing and processing, like computer vision, logic, reasoning, and so on.

Normally, those systems can’t talk to one another, so my lab is creating software that acts as an intermediary, and we’re testing it on a robot called Cindy. Using this software, she can understand plain language and use context to determine what information she’ll need to complete a simple task, like finding a box in a room.

It sounds simple, but underneath, it’s incredibly complex. To get there, we’ve needed to understand what it takes to make a mind—something that can perceive, think, and act.”

—Matthias Scheutz, associate professor of computer science and director of the Human-Robot Interaction Laboratory at the School of Engineering

Diagnosing Fetal Genetic Disease

“For a long time, only two procedures have been available to diagnose fetal genetic disorders. Both involve using a needle to remove cells from either the placenta or amniotic fluid that surrounds the fetus. By analyzing genetic material from those cells, you can tell for sure whether the fetus has a disorder like Down syndrome.

Unfortunately, these procedures come with a risk of miscarriage, so for many years, my lab has been interested in noninvasive ways to get the same genetic information. One thing we’ve worked on is using free-floating DNA—from the placenta—that is released into the mother’s blood. We have helped to develop a simple blood test to extract that DNA without causing any harm to the fetus. Within the past year, this test has already been incorporated into clinical care.

It’s incredibly useful, but it still doesn’t solve the problem of what to do after diagnosis. For me and my laboratory group, the challenge is how to treat genetic conditions while the fetus is in the womb.

Right now, we’re working with mice that have a genetic equivalent of human Down syndrome, and we’re using them to identify drugs that can help improve learning and memory in their newborns. We’re years away from clinical trials, but I think it could eventually make a big difference for the lives of children with genetic disorders. In a way, the work we do is sort of the final frontier. In medicine, we put a lot of emphasis on the end of life, but we don’t have nearly as many resources to prevent problems at the very beginning of life.”

—Diana Bianchi, executive director, Mother-Infant Research Institute, Tufts Medical Center

Stopping Lyme Disease at the Source

“There’s no magic bullet against Lyme disease, but we may be able to target its source. The bacterium that causes the disease, Borrelia burgdorferi, is carried by small animals like mice. As tiny larval ticks feed on the mice, they pick up the infection, which they spread to other animals when they feed again later in life. Right now, there’s no human vaccine, but there is one that works on mice. If we can vaccinate them against Lyme, we may be able to reduce the number of infected ticks.

Professor Linden Hu at Tufts Medical Center and I are testing this idea right now, using special bait containing the vaccine. As mice eat the bait, they become immunized to Lyme, so they won’t infect ticks. We know this method works in the lab. We’re now working on getting the vaccine approved by the USDA, and should be able to move into field trials in about two years.”

—Sam Telford, professor of infectious diseases, Cummings School of Veterinary Medicine

  © 2012 Tufts University Tufts Publications, 80 George St., Medford, MA 02155