Improving post-disaster response
UB receives $2.5 million grant to apply information fusion to emergencies

By ELLEN GOLDBAUM
Contributing Editor

Minutes after the first hijacked plane hit the World Trade Center's north tower on Sept. 11, information began to flow from the site to police, firefighters and other emergency personnel.

The situation and information flow became more complex when a second plane flew into the WTC's south tower, and even more complicated when both towers subsequently collapsed.

The science of efficiently organizing and interpreting massive amounts of such information—a relatively new field called information fusion that originated from military applications—is for the first time being applied to manmade and natural disasters by UB researchers.

"The technology exists for us to do better than we are doing now in responding to these kinds of crisis situations," said James Llinas, research professor of industrial engineering who is director of UB's Center for Multisource Information Fusion and a founder of the field.

"By helping us to both exploit and compress all the disparate information emanating from sensor systems and human reporting, information fusion will assist us in forming more effective assessments, and yielding improved decision-making in order to save lives and minimize damage to property."

The ultimate goal of the work being led by Llinas is new software tools to make the emergency response more effective by "fusing" the many channels of information that begin flowing following a major disaster. The integrated, automated fusion and decision-making software would be embedded in the urban emergency-management and crisis-management systems.

The work is funded by a five-year, $2.5 million grant awarded earlier this year by the Air Force Office of Scientific Research to the Calspan-UB Research Center (CUBRC).

Llinas said the grant takes advantage of UB's leadership in information fusion—the focus of the Center for Multisource Information Fusion, the only university center in the U.S. dedicated to the field of information fusion for multidisciplinary applications—and disaster mitigation—the focus of the Multidisciplinary Center for Earthquake Engineering Research, the National Science Foundation-funded national center of excellence in advanced technology applications that seeks the reduction of earthquake damage and losses.

"The existing high quality of our research in information fusion and earthquake engineering enables us to respond quickly to new demands, applying our research to solve critical problems," said Provost Elizabeth D. Capaldi. "We are proud of our ability to reach in new directions because of the excellence of our basic programs."

The University of Virginia at Charlottesville is UB's academic partner on the grant, leading some of the modeling work of post-disaster event dynamics.

UB's industrial partners are the leaders in the information fusion field, including Veridian Corp. (formerly Calspan); Alphatech Inc. of Burlington, Mass.; Orincon Corp. of San Diego, and IET Inc. of Arlington, Va.

Mark Karwan, dean of the School of Engineering and Applied Sciences, said the UB researchers, "who previously demonstrated considerable success in their own right, have combined their expertise to focus on the cutting-edge of highly meaningful research, made all the more critical in light of the events of Sept. 11."

Llinas said that in addition to manmade disasters such as terrorist attacks, chemical spills and biological attacks, the research will be relevant to information flow in natural disasters, particularly earthquakes; emergency-relief services during humanitarian missions; peace-enforcement operations; search-and-rescue missions; aeromedical evacuation and mercy missions, and traditional military situations.

The objectives are to improve the immediate response to disasters with particular emphasis on assessing damage, search and rescue, sheltering and evacuation, and resource management.

Immediately after a major manmade or natural disaster, Llinas explained, much of the information that emergency personnel receive describes the state of urban "lifeline infrastructure," collapsed buildings and damaged roadways, bridges, power lines, water-distribution systems, utilities and transportation systems. That information comes into emergency-response offices at the local, state and federal level in various ways—through satellite images, sensor systems embedded in the infrastructure, police reports, property owners and other individuals.

"The role of information fusion here is to develop new, improved methods for processing observational data in order to generate improved estimates of the state of urban lifeline infrastructure components and networks so that emergency professionals can make the best estimates to support their response decisions," said Llinas.

Under the grant, a new and elaborate software program for route optimization—the fastest and best way for vehicles to get from point A to point B—that was developed by UB postdoctoral associate William Frank, now will be adapted to "dynamic route optimization," that is, to routes that have been disturbed by an earthquake, a terrorist attack or some other natural or manmade disturbance.

Llinas noted that what makes these kinds of programs particularly challenging is that damage to roads and infrastructure does not remain static. For example, after a building collapses and the emergency effort begins, other nearby buildings may buckle and collapse, or in the case of an earthquake, aftershocks may cause additional roads to become impassable.

"What we want to do here is to fuse all incoming information to create, for example, the best estimate of the status of road links and feed that into the route optimization process," he said.

Information fusion emerged during the Cold War as a way of streamlining intelligence, surveillance and reconnaissance information gathered by the military, Llinas explained.

"Fusion, it was realized, would result in a single merged lane of traffic, so to speak, instead of many lanes of traffic proceeding separately. At the same time, and even more importantly, fusion would result in processing that would have higher information content and less uncertainty."

CUBRC funded initial research instrumental in acquiring the grant, and is providing technical consultation in the area of chemical and biological defense. A not-for-profit corporation, CUBRC brings together engineers and scientists from Veridian Corp. with UB faculty and students to pursue research-and-development programs that foster technology transfer.

Llinas, principal investigator on the grant, has been honored by the International Society for Information Fusion for lifetime contributions to the field. He is a former executive engineer at Veridian Inc., and co-author of the first information fusion textbook.

UB co-investigators on the grant include George C. Lee, director of MCEER and Samuel P. Capen Professor of Engineering at UB, and Ron Eguchi, president of ImageCat and an MCEER-funded researcher.

Also, Thenkurussi Kesavadas, assistant professor of mechanical and aerospace engineering; Li Lin, associate professor of industrial engineering; Christopher Rump, assistant professor of industrial engineering; Peter Scott, associate professor of computer science and engineering; Aidong Zhang, associate professor of computer science and engineering; Galina Rogova, adjunct professor of industrial engineering, and William Frank, a post-doctoral research scientist.