Sonia Chernova, assistant professor of computer science and robotics engineering at Worcester Polytechnic Institute (WPI), has received a five-year, $500,000 CAREER Award from the National Science Foundation (NSF) to conduct research aimed at paving the way for general purpose robots that can work effectively and productively alongside people in everyday settings. Chernova was one of two WPI faculty members to receive the CAREER Award, the most prestigious NSF award for faculty members early in their careers as researchers and educators, this year and is one of 18 current WPI faculty members who have won this honor.
"We congratulate Professor Chernova on earning this exceptional award," said WPI Provost Eric Overstr-m. "With this grant, she will undertake important work with the potential to bring about significant advances in her field and profound improvements in our quality of life."
General purpose robots must able to adapt easily to take on new jobs and respond to the changing needs of their human partners-for example, performing routine housekeeping tasks to help the elderly continue living in their own homes or enabling small manufacturers to quickly gear up to make new products without having to employ a staff of programmers. The need to accelerate the development of such robots was an important element of the robotics research roadmap presented to Congress in 2009 by the Computing Community Consortium and the National Robotics Initiative announced by President Obama in 2011.
Chernova argues that to realize this vision, the way robots are taught to perform tasks has to fundamentally change. "Currently, we attempt to preprogram all of the knowledge a robot needs to operate in our world," she says. "In this way, we've been successful at developing robots to work in constrained environments, such as assembly lines. But we have not been so successful at getting robots to work in natural, unconstrained environments, such as homes. To enable robots to adapt to new and changing environments, we need a way to customize a robot's behavior on the fly, without the need for skilled programmers."
With her CAREER Award, Chernova will attempt to determine what robots need to know to perform various tasks and how everyday people, with no understanding of programming or even how robots work, can help them gain that knowledge. To accelerate this research, she'll draw on the power of crowd-sourcing-creating a web-based testing environment that will enable her to gather data from thousands of Internet users. Designed as a game in which players advance through progressively more difficult tasks-first with a simulated robot and later, after they've achieved sufficient proficiency, with a physical robot-the system will ask users to show a robot how to perform tasks in a home (tidying up a room, for example) and in a small-scale manufacturing plant.
Unlike typical robots, the machines used in Chernova's research will be capable of active learning, which means that they will be able to ask questions. Since her aim is to develop robots that operate in proximity with people, it makes sense, she says, for the robots to ask people for help. In particular, the robots will be able to ask which features of objects or of the environment are relevant to a particular task-for example, whether color is important in determining whether an object on a table is a book or a magazine.
"People are excellent at ruling out irrelevant features and knowing what is important and what isn't," Chernova says. "Developing similar capabilities for robots is critical to enable them to operate effectively in our complex world."
Clarification urgently needed for detected signal of semaglutide-linked suicidal ideation