Can robots prevent deadly stampedes?

GREAT MINDS: Haibo He, professor of engineering at the University of Rhode Island, left, and one of his Ph.D. students, Xiangnan Zhong, discuss the research they are doing to produce robots to help humans in emergency situations such as stampedes. / PBN PHOTO/ MICHAEL SALERNO
GREAT MINDS: Haibo He, professor of engineering at the University of Rhode Island, left, and one of his Ph.D. students, Xiangnan Zhong, discuss the research they are doing to produce robots to help humans in emergency situations such as stampedes. / PBN PHOTO/ MICHAEL SALERNO

A stampede in September killed more than 700 people outside the holy city of Mecca, in Saudi Arabia. A fire in October killed more than 32 people inside a Bucharest nightclub in Romania. A panicked crowd last December killed at least 36 people during a New Year’s celebration in Shanghai.

These tragic events represent only a portion of the deadly disasters that claimed thousands of lives this past year and are among a growing number of incidents that have inspired one University of Rhode Island professor to develop robots that could reduce the number of casualties during emergency situations.

“The idea was from a few years back when my colleague and I were discussing the number of stampedes at many public events that happen either due to fires or alarmed [crowds],” said Haibo He, URI professor of electrical engineering. “We thought, what type of technology could help to prevent or minimize this type of damage?”

He and his colleague, Yi Guo, an associate professor of electrical and computer engineering at Stevens Institute of Technology in New Jersey, quickly turned their attention to research. The duo produced an idea to develop robots that could mimic human behavior, adapt to changing environments and help guide people to safety.

- Advertisement -

“If we have a shopping mall and there is an alarm, people may try to rush to the nearest exit, but because there’s only one door it will create a slower evacuation situation,” He said. “But if we can deploy multiple robots and guide people to different exits or doors that may be farther away, we can optimize the human flow.”

A John Hopkins School of Medicine study suggests at least 215 stampedes worldwide happened between 1980 and 2007, resulting in more than 7,000 deaths and twice as many injuries. To bring the example home, in 2003, The Station nightclub fire in West Warwick killed 100 people and injured more than 200 others, marking one of New England’s deadliest disasters. Since people tried to exit through the front door once the fire started, it quickly became blocked, resulting in a number of deaths.

The John Hopkins research suggests that when people are in a crowd of about 10 people per one square meter, they tend to move as one and “ignore alternate exists, accelerating the possibility of disaster,” according to John Hopkins magazine.

He’s robots, which haven’t yet reached prototype, would theoretically stay cool under these types of pressure and make the most optimal decision to guide people away from an emergency.

To work, however, He and Guo will have to program the unmanned robots to understand and mimic human behavior, know how to adapt to any given emergency and use real-time data to make quick decisions in order to optimize human flow and safety.

“They will control themselves, so we have to develop the intelligence,” He said.

The research group has grown since its inception and now comprises He, Guo, URI psychology professor Charles E. Collyer, three doctorate students from URI and another from Stevens Institute.

Collyer specializes in harm reduction and says this project is likely a leading example of the more “sophisticated interactions between robots and people,” which makes it a good candidate for his field of psychology.

“This is an example of how engineering and psychology interface quite often,” Collyer said. “Engineering is often dealing with behavioral issues in addition to technology issues, and this kind of collaborations make sense.”

In October, the team received a $282,000 grant from the National Science Foundation and is now presenting the research at conferences and meetings around the country. He says a critical part of the three-year project will be to establish the human-to-robot interaction, as the robots will need to know how to identify patterns based on human behavior and any given environment.

For instance, He says, at shopping malls, research shows the highest volume of people entering from 8 a.m. to 9 a.m., most people eating at the food court from 11 a.m. to 2 p.m. and most people exiting from 4 p.m. to 5 p.m. The pattern, although simple, would tell a robot where the highest density of humans would be at any given time should an emergency arise.

Preliminary results thus far have been “promising,” He said, and although it’s not a part of the original proposal, He believes the robots could be used in the future during nonemergency scenarios, whether it be for sporting events, concerts, amusement parks or even Black Friday, when human flow could be optimized.

“We can use this type of technique to optimize human movement and reduce the waiting time and increase movement efficiency,” He said. •

No posts to display