URI_Research_Magazine_Momentum_Spring_2016_Melissa-McCarthy

Not just any brain, but a self-learning, self-adapting brain, which if successful, will allow human-sized robots to indirectly guide the behavior of humans in emergency situations.

At the University of Rhode Island (URI), an engineering professor is embarking on what may seem like a futuristic endeavor to find – or rather, create – a robot a brain. And not just any brain, but a self- learning, self-adapting brain, which if successful, will allow human-sized robots to indirectly guide the behavior of humans in emergency situations. Haibo He, the Robert Haas Endowed Chair Professor of electrical engineering at URI is working on the computational intelligence of a human-robot interaction project through a $282,000 grant from the National Science Foundation (NSF). The funding will allow He to develop learning-based control algorithms designed to create intuitively thinking brains for robots. The National Robotics Initiative, launched by President Barack Obama, is on the forefront of developing science. Along this line, He’s robotics research has a targeted application: improving human evacuations in emergency situations at congested, contained events, such as games or concerts. “If something happened and people try to exit the area as quickly as possible, that could cause a stampede,” He says. “Currently, there is little way to control the behavior of scared crowds, which can cause disasters.” He envisions a world where rather than sending more humans into dangerous situations, robots could be deployed to assist evacuations. His

Professor Charles Collyer to collect and analyze data on human movement from surveillance cameras. He is additionally working with Associate Professor Yi Guo at the Stevens Institute of Technology in New Jersey who is developing the physical robots. Their research team has also grown to include doctorate students from both URI and the Stevens Institute. “Robots are just robots,” He says. “It’s a mechanical, physical part. How to make it alive, how to make it intelligent – that is what we are working on.” Though the project is just beginning to gain momentum, He and his team have already collected interesting preliminary results that indicate this human-robot interaction could dramatically impact the safety and efficiency of human evacuations. He hopes to apply the results of his research to URI and the surrounding community. “Hopefully by the end of the project we can demonstrate the entire concept in a controlled environment such as a student dorm,” He says. When He presented his preliminary results around the globe, he received positive reactions from different entities, such as fire departments, and emergency management agencies, among others. “When they saw my presentation and they saw these results they were

robots would be able to anticipate the movements of humans, evaluate changes in the environment in real- time, and make educated decisions on how to evacuate people as efficiently and safely as possible. “In our current preliminary study, we can evacuate maybe 50 to 100 more people compared to the existing approach in a simulated environment in our lab,” He says. “But even one more person is a huge benefit because we are talking about saving lives.” He’s robotics-assisted evacuation project was inspired by two events that on the surface may seem quite different, but in which the professor saw natural similarities. Before applying for the NSF grant, He was shocked by news of a stampede at a New Year’s Eve celebration in 2014 in Shanghai that killed 36 people. In the same year, an article published in the Journal of the Royal Society Interface ended up in his lap that illustrated how farmers used dogs to herd their flocks of sheep. “They just placed a few dogs in key locations to ensure the whole flock of sheep moved in certain directions,” He says. “Kind of like our robots in these situations.” The NSF grant will allow He to work collaboratively with professionals who can tackle the other two parts of the project’s equation: understanding human behavior and building a physical robot on which to test He’s brain. He has partnered with URI psychology

Page 30 | The University of Rhode Island { momentum: Research & Innovation }