Smarter buildings
by Kathleen Ricker

They're simple, generate what little power they need, and can be produced for pennies on the dollar. But NCSA researchers Peter Bajcsy and Rob Kooper, with the help of two students, Miles Johnson and Kyaw Soe, hope that radio frequency identification (RFID) tags can be used to save lives and property--and perhaps even, in the future, prevent minor hazards from developing into full-scale disasters. Their project is one of six to receive a TRECC Accelerator Program Award, which provides them with a moderate infusion of funds to make progress toward a prototype. They also received funding for their project from NCASSR in 2004.

Prototype setup of robot-mounted
active sensor with configuration
of RFID tags.

Most people come in contact with RFID sensors every day. Access cards increasingly contain RFID sensors which transmit signals to readers that unlock hotel and office building entrances. A lost pet may be reunited with his owner if a veterinarian, using a scanner, finds an RFID tag under the animal's skin that provides identifying information.

RFID sensors are powered by induction. In the case of the access cards, when the access card is held closely to the reader mounted next to the door, a small electric charge will be generated that causes an RFID sensor to transmit a signal to the computer that controls the door. The RFID tags on which Bajcsy and hist team's research focuses, while smaller and more flexible, are still rugged, reliable, and fault tolerant and work on the same principle. "The biggest advantage is that you don't need any batteries," says Kooper.

This advantage is especially critical, considering what Bajcsy and his team are envisioning: entire buildings in whose floors RFID tags are embedded, triggered by a mobile, remotely-controlled "reader"--an active sensor mounted on a robot. The robot rolls over the RFID tags, providing its location partly through an algorithm developed by Bajcsy that calculates the probability of the robot's location based on the location of the tags it has triggered. The active sensor synchronizes with wired and wireless cameras throughout the environment and sends back more sophisticated information--e.g., the temperature of the room, or hazardous chemicals that might be in the air, providing remotely located humans with an idea not only of where the crisis is unfolding, but also what kind of emergency it is and how it might be handled.

"If you place the active sensor in a particular location," says Bacjsy, "the robot scans the room and says, there's a fire in that corner over there. How does the robot know? The RFID tags take care of that. The RFID tag can tell me, cheaply, where I am, and then I can use the more expensive active sensors to tell me more about the fire." Bajcsy and his team have even gone one step farther, simulating hazard containment: they attached a hair dryer to the robot carrying the active sensor and had it extinguish candles burning in a test environment. In the same way, a robot could be equipped with a small fire extinguisher to handle small fires before they became larger. “Even if the hazard is not completely contained, its rate of increase can be slowed,” says Bajcsy. “A few minutes—even a few seconds—can make a huge difference.”

Because RFIDs have limited reading range and sensing capabilities, TRECC’s infusion of funding was especially important to Bajcsy and his team as they explored the properties of the RFIDs and how they could best be made to work with active sensors. By running a single-antenna active sensor mounted on a small robot over various RFID configurations laid out on the floor, they found that RFIDs could be placed a maximum of 25 inches apart on the ground. They also found that it made no difference whether the active sensor traveled rapidly or slowly over the RFIDs—meaning that the active sensor could cover ground more quickly. They concluded that wood, paper, and plastic objects up to four inches in thickness did not interfere with readings, but metal objects did. They found that placing more than 10-15 sensors in a small space would create a bottleneck that would make localizing the active sensor more difficult. And they found that the optimal RFID shape for their purposes was long and slender—about six inches.

Making a space itself hazard-aware is much more precise than systems which alert emergency services that there may be a fire in a residence or an office building, Bacjsy points out. When a signal is sent to a fire department, responders don't necessarily know where the fire is, whether anyone is at home, or even whether the emergency is real--or whether it's a false alarm. But fighting fires, says Bajcsy, is only one of many uses for his system. He envisions applications for remote environments rarely occupied by humans, like observation stations, or too dangerous, like nuclear reactors. And he also envisions hazard-aware living spaces that could help caregivers and emergency responders track elderly and disabled people who may live alone.

TRECC’s funding played a crucial part in helping Bajcsy, Kooper, Johnson and Soe to complete concept validation and create a robust prototype suitable for technology transfer or commercialization. “By having the TRECC funding, we could explore new technologies that could eventually lead to a robust system, so that someone who was interested in developing a shrink-wrapped solution from it would have the knowledge and expertise to go about doing it,” says Bajcsy. Indeed, Bajcsy and Kooper are now investigating such opportunities—and they’re beginning to attract interest.