Research

Safety in numbers: Mobile robots could save lives during emergency evacuations

Researchers receive $1.5 million National Science Foundation grant to develop first-of-its-kind robot-assisted evacuation system

Alan Wagner, assistant professor of aerospace engineering, and graduate student Sagar Lakhmani adjust a pair of Braccio robotic arms on the latest generation of emergency evacuation robots in the Penn State Robot Ethics and Aerial Vehicles Lab. Credit: Penn State. Creative Commons

UNIVERSITY PARK, Pa. — As recent global events have shown, knowing how to evacuate during an emergency is becoming a necessary component at concerts, movies, churches and even for children attending school. Many of these emergencies require people to evacuate quickly and to make quick decisions; however, when people rely on default decision making, they could be putting themselves or others at risk.

To address the unique challenges that emergency evacuation scenarios present, Alan Wagner, assistant professor of aerospace engineering, and a team of researchers propose to use a coordinated collection of mobile robots as authority figures to direct evacuees for rapid, orderly and safe evacuations.

The success of their research efforts, which will be funded by a recently awarded, four-year, approximately $1.5 million grant from the National Science Foundation, could broadly impact the public’s safety and security — not only potentially saving lives but also help with improving crowd control, preventing stampedes and keeping families together.

“Because robots are becoming more and more pervasive on city streets, in hospitals and in hotels, they have the potential to serve as a type of instantaneous first responder, coming to assist immediately after an emergency occurs,” said Wagner, principal investigator (PI). “When compared with traditional emergency infrastructure technologies such as fire alarms and smoke detectors, mobile robots can achieve better situational awareness and, hopefully, use this information to expedite evacuation and enhance safety.”

He explains further that mobile robots could also be used in risky and life-threatening situations such as chemical spills or active-shooter scenarios, which present dangers to human first responders. Beyond that, the researchers hope that their work could extend to areas involving cooperative robot teams that are embodied in an uncertain and dynamic physical world and need to actively interact with humans, such as the battlefield, law enforcement, factories and urban transportation systems.

Substantial work has been done on the mathematical modeling of large-scale evacuations of a populace, but the study of robot-assisted evacuation is only very recent. Simulated evacuations have clearly demonstrated that robots are able to speed the evacuation process, but these simulations only considered a single robot and not multiple.

Wagner, along with co-principal investigators Minghui Zhu, assistant professor of electrical engineering at Penn State, and Hai Lin, associate professor of electrical engineering at the University of Notre Dame, intend to derive systematic methods of designing coordinated robot decision making and motion planning in human-crowded environments to achieve an efficient evacuation. They will also investigate the human-robot interaction issues associated with evacuation through real human-robotic experimental studies and evaluate the effectiveness of their theoretical and experimental results by creating a system to be used in field tests.

“Our overarching goal is to create the first system of this kind — one that is scalable and customizable — for evacuations of buildings, schools and concerts, whereby we can coordinate a collection of robots so that, when an emergency alarm is generated, the robots will be deployed immediately to critical navigation points within the environment. Once there, they will guide the flow of evacuees and create an efficient evacuation process,” said Wagner. “The key novel feature of our approach will be the systematic integration of human crowd simulation models with methods from control theory, game theory, machine learning, optimization and human-robot interaction.”

The robots will not be designed to lead a human group or physically assist humans to move. Instead, when deployed and strategically positioned, the robots will act as “traffic cops” directing people toward the exit minimizing evacuation time, while keeping crowd density below a hazardous level. Should a pathway become blocked or congested, the robots would reposition themselves and dynamically redirect evacuees away from congestion points and toward an open exit. Once the flow of evacuees subsides, the robots would move up and down hallways, verbally providing information to evacuees.

The robots would also be designed to estimate evacuees’ psychological states, such as stress and panic level, using onboard cameras, while also sending streaming video to human rescuers.

Because emergency situations and environments can change dynamically and quickly, and the operating environments for mobile robots may be unstructured and crowded, the robots will be developed to use synthesized motion planners based on integrated sampling-based algorithms and game theory to avoid high-density areas and collisions.

The robots will also be designed to communicate directly with evacuees or individuals who are sheltering in place in an attempt to comfort them.

“Although it’s a challenging human-robot interaction problem, the robots must be able to effectively communicate with evacuees and direct them to safety, while also maintaining their attention and preventing congestion,” said Wagner. “The robots must communicate these directions to people that have little or no experience with robots, using both verbal and nonverbal communication, and it is currently unknown whether and under what conditions people will follow a robot’s emergency evacuation instructions.”

To validate their models and demonstrate the effectiveness of their proposed design, the researchers plan to conduct human-robot interaction studies in the final year of the proposal, using human subjects during an emergency evacuation. Buildings on the University Park campus will be used as test locations, and approximately 100 participants will be recruited from the University, State College and the surrounding communities.

Wagner, who is also affiliated with the Penn State Rock Ethics Institute, intends to incorporate this research into a series of ethical case studies related to the project for future engineering courses. Along with Zhu, Wagner also plans to develop a youth summer day camp that will engage students interested in science, technology, engineering and mathematics by introducing them to robotics-related applications, including emergency evacuation.

Last Updated January 21, 2019