Penn Engineering Establishes Intel Center for Wireless Autonomous Systems

These autonomous robots work together to efficiently map unknown spaces or locate targets. Wireless robot-to-robot communication is key to coordinating their behavior, but new approaches are necessary to circumvent challenges inherent to cellular and Wi-Fi connections.

Penn Engineering Establishes Intel Center for Wireless Autonomous Systems

The University of Pennsylvania’s School of Engineering and Applied Science has established the Intel Center for Wireless Autonomous Systems. The research group, made possible by a three-year, $1.5 million gift from Intel, will investigate how robots and other machines can best wirelessly communicate with each other in high-stakes situations.

Increasing the reliability of wireless networks is a necessary step in developing a variety of applications, such as swarms of flying robots that can deliver packages or “smart” cities that coordinate fleets of autonomous vehicles.

“Receiving a gift of this magnitude from an industry partner is quite rare,” said Alejandro Ribeiro, the Center’s director and Rosenbluth Associate Professor in the Department of Electrical and Systems Engineering (ESE).

The Center is a part of a series of Intel Labs-sponsored academic research partnerships. Each Intel Science and Technology Center is focused on a different emerging technology, in this case, “Context Aware Wireless Networking for Autonomous Systems.”

“The Intel Science and Technology Center on Wireless for Autonomous Systems at the University of Pennsylvania will help address some of today’s biggest issues surrounding unreliable wireless connectivity,” said Vida Ilderem, vice president and director of Wireless Communication Research at Intel. “This requires fundamental advances in the co-design of wireless and control systems to enhance key functions of sensing, computing, cooperation, and decision making. We believe our multidisciplinary research will improve wireless service levels dramatically.”

Standard cellular and Wi-Fi connections are sufficient for consumer electronics like phones and laptops, but they constantly contend with the inherent limitations of wireless media. Machines in wireless autonomous systems need higher reliability and speed than those standards can currently provide.

“When your cellphone drops a call, it’s a problem of physics,” said Ribeiro. “The electromagnetic waves that carry that information are bouncing off walls, objects and people, and sometimes those waves interfere with each other and cancel each other out. Losing a call or a slow-loading webpage is annoying, but when it comes to the systems that control groups of flying robots or self-driving cars, it can be dangerous.”

Getting around these physical limitations is possible, but costly: Adding power or redundant signals to cellular and Wi-Fi networks drains batteries and increases wireless traffic in already crowded bandwidths. “Context aware” wireless networks would be better able to manage such tradeoffs.

“With a finite amount of resources to utilize, we need to design communications that are aware of and can adapt to the state of the system,” Ribeiro said. “That way, you can generate extra reliability only when you really need it.”

Context aware tradeoffs can take many forms. A team of flying robots might be programed to accept the possibility of dropping out of communication when they are hundreds of feet apart, but expend more power on transmission redundancy as they get closer together. It might also mean physically rearranging their positions to avoid patches of communication interference.

Mobile autonomous systems are made even more complicated by human users. Highly reliable and context aware wireless networks are even more important when considering self-driving cars, as they would be able to manage entire cities of autonomous vehicles that need to talk to one another to optimize traffic patterns.

Beyond these mobile systems, the Center’s research also has applications in industrial settings, where cabling between stationary assembly line robots can be a substantial percentage of a factory’s cost.

Other Penn-Engineering-affiliated members of the Center include Firooz Aflatouni, Skirkanich Assistant Professor in ESE, Vijay Kumar, Nemirovsky Family Dean of Penn Engineering and professor in Mechanical Engineering and Applied Mechanics, Dan Lee, UPS Foundation Professor in Transportation in ESE and director of the GRASP lab, Rahul Mangharam, associate professor in ESE and Penn’s director of Mobility21, a transportation research center, and George Pappas, Joseph Moore Professor and chair of ESE.

Dina Katabi of the Massachusetts Institute of Technology, where she is the Andrew & Erna Viterbi Professor of Electrical Engineering and Computer Science and the director of the MIT’s Center for Wireless Networks and Mobile Computing, is also a member of the Center.