Radke Receives NSF CAREER Award

Richard Radke, assistant professor of electrical, computer, and systems engineering at Rensselaer, has been awarded a Faculty Early Career Development Award (CAREER) from the National Science Foundation (NSF). Radke is among 22 Rensselaer faculty members who have received the award in the past four years.

Thomas Griffin

The CAREER Award provides a grant of $400,000 over five years, and is the most prestigious honor presented to junior faculty. Radke will use the award to fund a graduate student to assist him in developing a new framework for “distributed computer vision.” This mathematical system, he says, will someday allow thousands of video cameras to automatically work together to map distant or inhospitable areas, or track potential enemies or criminals. Integrating the many separate streams of information that individual cameras collect into a cohesive picture of the area in question is a challenging research question, says Radke.

This mathematical system will someday allow thousands of video cameras to automatically work together to map distant or inhospitable areas, or track potential enemies or criminals. —Richard Radke—

Picture a typical security camera system, the kind you might see in a large airport or office building. The images gathered by each camera are sent to a bank of video screens — that a human being (or multiple human beings) has to monitor — hardly a high tech operation.

“There is nothing automated about that kind of system,” says Radke. “Instead, I want the cameras to work together locally to create a global picture. To do that, I have to establish a chain of conversations between nearby cameras.”

Radke is working on an algorithm that will allow each camera to communicate with its neighbors, comparing landmarks and other features to determine its location and help build one master map.

Radke will start close to home, by using cameras that he will attach to various Rensselaer campus buildings. He plans to begin installing the cameras this fall.

“Eventually, they could be outfitted with optical antennae that allow large amounts of data to be shared quickly,” says Radke. “My first goal is for the cameras to automatically figure out where they are, and where they’re pointed. Then we’ll work on the view synthesis problem.”

The movable cameras should be able to see the public behavior of pedestrians and vehicles, says Radke, but they won’t capture enough detail to see into windows or identify specific individuals.

Radke plans to make the cameras remotely steerable, and hopes to involve children who visit Troy’s Junior Museum and students who participate in Rensselaer’s Questar III program. Questar III provides hands-on, college-level experience to high school students considering a college major in math, engineering, information technology, or science.

“Having kids at the museum randomly steering the cameras that are located on campus will definitely make integrating the data more challenging,” he says. “But that’s the point of the exercise.”