Collaboration with civil and environmental engineering looks at floor sensor data -Silicon Valley Campus - Carnegie Mellon University

Collaboration with civil and environmental engineering looks at floor sensor data -Silicon Valley Campus - Carnegie Mellon University

Tuesday, March 25, 2014

Collaboration with civil and environmental engineering looks at floor sensor data

Originally published on the CMU Civil & Environmental Engineering site 
By Laura Pacilio

Soon, banks may not need cameras to spot criminals. That’s because CEE Assistant Professor Hae Young Noh is developing ways to record and track people’s footsteps. The project, which is a collaboration with associate research professor Pei Zhang (ECE - CMU Silicon Valley) and associate professor Lin Zhang (Tsinghua University), involves analyzing the data collected from floor sensors. “Usually people don’t think about it,” she says, “how much information they are giving out by just walking around.”  

The sensors pick up the vibration waves that pass through the floor when someone takes a step. By examining these these waves, Noh can figure out how fast someone is moving, their weight, and even their shoe type. “The waveform actually looks different depending on how you walk, whether you’re wearing high heels, etc.,” she says. For instance, high heels produce higher frequency vibration waves than tennis shoes. When a person is turning, they place more weight on one foot, producing a footstep wave that is noticeably higher than the others.

The sensors can also be used to find people. If you put three sensors in various places along the floor, footstep waves will arrive at each sensor at different times with different amplitudes. The time and amplitude differences can be used to pinpoint a person’s exact location. Because of this, many businesses have expressed interest in using this technology for security purposes. Noh says that retailers could also combine sensor data with the information from clothing tags to determine what items customers are actually bringing into the dressing room.  

It’s even possible to use the sensors with machine learning algorithms that can recognize specific walking patterns and flag people who may be sick, nervous, or lost. This could help businesses identify new customers who need assistance and nursing homes identify patients who need medical attention. This kind of tracking depends on the floor type and the number of people in a space—eventually, individual footstep waves begin to blend together—but Noh says that with enough sensors, she could track hundreds of people.  

In very large areas like shopping malls, the sensors use the floor’s general vibration level to estimate the number of people and what kinds of activities they’re doing. Because these factors greatly influence the amount of electricity and heat used in a building, the sensors are also an excellent, nonintrusive way to monitor energy consumption. 

With the help of PhD students Chandrayee Basu (CEE) and Shijia Pan (ECE) and undergraduate Amelie Marie Bonde (CS ’14) — who are co-advised by Noh and Zhang — Noh has already published a paper describing this work. Over the next few years, she plans to conduct lab experiments to refine her understanding of sensor data and is also interested in employing this technology in other contexts. “Vibration analysis can be applied to floors, buildings, trains, and different things. Vibration is everywhere,” she says. One new project even involves placing sensors inside of clothing so that they can analyze the vibrations of people’s muscles.  

Because of their many uses, Noh anticipates that sensors will eventually be a standard feature in a myriad of spaces.  That means vibrations are quite literally the waves of the future.