ECE Makes Major Advances in Cyber-Physical Systems


September 13, 2012

Assistant Research Professor Anthony Rowe places a $300 plastic flying toy drone on the floor inside a parking garage in Carnegie Mellon’s Collaborative Innovation Center. The drone has four propellers and measures about a foot-and-a-half wide. Camera lenses on the bottom and front send images to a human controller on the ground.

By design, an iPhone controls the drone — tilt the phone forward and the drone flies forward; tilt it back and the drone flies backward — but Rowe and his students are retrofitting it to fly without a pilot. They also want it to communicate with other devices across a wireless network, much like cell phones already do.

Ask Rowe what the point is and he might tell you to imagine a scenario like the aftermath of Hurricane Katrina. Landlines are down, cell phones won’t work, roads are jammed and people need to be saved. Rescue workers arrive on the scene and release small flying drones from their backpacks. Once airborne, the drones self-arrange into a grid above the city and start relaying visual information to rescue workers on the ground. At the same time, they set up an impromptu communications network that restores cell phone usage across the city, providing a lifeline for those in danger.

Unlike helicopters and state-of-the-art military drones, these easily deployable minidrones can blanket and monitor an entire citywide airspace without landing pads, fossil fuels or pilots. And at less than $1,000 per unit, they’re an affordable alternative to the $4.5 million-per-unit piloted military drones actually used in Hurricane Katrina’s aftermath.

Welcome to the world of cyber-physical systems — one of ECE’s most rapidly growing fields of interdisciplinary research and development.

Cyber-physical systems (CPS) result from adding computational and/or communications elements to previously passive physical tools. “To me, cyber-physical systems refer to the embedding of sensing, communications and computing into physical spaces,” said ECE Assistant Professor Bruno Sinopoli. “This makes physical spaces smarter, more comfortable and more secure.”

One familiar example of a cyber-physical system is the cruise control function in cars. Once, speed regulation required constant monitoring by a human driver, but most new cars can regulate their own speed at the touch of a button. Likewise, CPS technology will allow Rowe’s drones to hover in formation and reconfigure themselves according to specified objectives without the need for direct human piloting.

Impressive as they sound, autonomous flying drones are Rowe’s least-developed project. He also created energy-monitoring applications for CMU’s FireFly Wireless Sensor Network, a low-cost, low-power hardware sensor platform that collects and transmits data in real-time. By attaching FireFly nodes to ordinary household appliances, Rowe can monitor and record household energy use in intimate detail and turn appliances on or off automatically.

“In the future, this technology and the data it collects will be vital in helping power-hungry appliances avoid operating at the same time, which will minimize expensive peak-usage periods,” Rowe said. If applied in households across an electrical grid, FireFly will increase efficiency, which will reduce the need for new power plants, help eliminate blackouts and curb the environmental impact of energy consumption.

Like Rowe’s modified flying drones, FireFly falls into the CPS category because it gives computational and communicational (i.e., cybernetic) abilities to physical hardware that previously had no cybernetic functions.

George Westinghouse Professor Raj Rajkumar, who co-directs the General Motors-Carnegie Mellon Autonomous Driving Collaborative Research Lab, believes cyberphysical technology will improve nearly every domain of our lives. One day, buildings will adjust their temperatures according to the weather forecast, or doctors will perform battlefield surgeries from hundreds of miles away. To convert these ideas into realities, Rajkumar and Rowe are collaborating with their colleagues at CMU and around the world to advance cyber-physical technology.

“Our goal is to lead a community across the country and around the world to build a case for a vision for CPS and lobby federal agencies to make funding available,” Rajkumar said. “Universities like Carnegie Mellon, the University of Illinois at Urbana-Champaign, the University of California at Berkeley, Case Western University and the University of Texas at Austin have banded together. This group works very closely with the National Science Foundation and has been successful in creating a five-year, $50 million-per-year program.”

Through his own work, Rajkumar aims to create cars that drive autonomously and communicate with each other, ultimately turning traffic into a cyber-physical system — the continuation of a process that began decades ago with cruise control and antilock brakes.

“If you embed a computer into something, it becomes a much smarter platform. For 80 or 90 years, the automobile was a purely mechanical platform. But in the past 10 to 15 years, computers have started getting in,” Rajkumar said. “Cruise control is really a computer that sees how fast you’re driving and what your speed setting is, and tries to catch up or slow down to that speed. That is the ‘cyber’ part of a cyber-physical system.”

Rajkumar envisions a world where cars not only drive themselves, but also communicate with each other to manage traffic flow and avoid collisions. Instead of relying on human drivers to stop at timed traffic signals (even when no opposing traffic is coming), cyber-physical cars will alert each other wirelessly, stopping or slowing down as needed. A car will become part of a cyber-physical communications network capable of considering every imaginable variable — traffic flow, oncoming cars, shortest routes, unexpected obstacles or road work — to transport passengers with maximum efficiency and safety.

That increased safety is the greatest selling point of autonomous driving.

“Humans get distracted all the time while driving, but computers are not emotional. They can be vigilant all the time,” Rajkumar said. “With the widespread implementation of autonomous driving, I can see the number of car deaths per year in the U.S. going from 35,000 to 3,500 to 350. The number may never reach zero, but, by and large, autonomous driving will be a huge win for society.”

If Rajkumar and his colleagues have their way, autonomous driving won’t be society’s only win. Technologies like Rowe’s drones will improve disaster response; sensors on buildings or bridges will warn engineers of impending structural dangers; and new advances in technology in the home will provide lifelong assistance for the elderly or those with disabilities. And that’s just the start.

Thanks to Carnegie Mellon researchers, the possibilities are endless.

Related People:

Anthony Rowe