Sensor network could aid mass transit

Human behavior can be observed and analyzed accurately by a complex sensor network, a system that ultimately could benefit public transportation, homeland security, and crime prevention, reports a research team at the Indiana University School of Informatics, Indianapolis, and two universities in Japan.

The researchers evaluated a distributed sensor network they deployed in the JR Kyoto subway station in Japan as part of the Digital City Surveillance Project. They blanketed the station with a sensor network of 28 wide-view cameras, and developed a system that "learned" from a station operator to recognize what people were doing in the concourse. The system then audibly told operators about events like overcrowding so operators could respond promptly. "The advantage of this system is that it doesn't require an expert to design it to recognize a certain kind of event. Station operators can do that as new situations arise," explains Karl F. MacDorman, associate professor of informatics.

MacDorman, an android science and robotics specialist, points out that sensor networks are becoming increasingly important in supporting interaction between humans and robots. "This is a large-scale practical system that incorporates learning. Normally, the system designer develops an explicit model of human beings and specific modules for recognized different kinds of behavior. Our system allows for the system to be trained by nontechnical station operators in an hour or two."

The wide-view video cameras used in the study combined flat and hyperbolic mirrors. Apart from public transportation, such systems can be used commercially to detect shoplifting and vandalism--and they could be employed for homeland security in areas such as bomb detection and monitoring of border crossings.

Sensor networks already are part of everyday life for many people. "There's not much you can do in, say, central London that doesn't show up on somebody's monitor," MacDorman points out. "All of the cameras just aren't linked together yet, and human operators are still relied on--but that's changing."

COPYRIGHT 2007 Society for the Advancement of Education
COPYRIGHT 2008 Gale, Cengage Learning