Naomi Ehrich Leonard's School of Collaboration

theory and its applications to autonomous underwater vehicles. Leonard's talent lies in creatively exploring solutions to control problems through combining diverse expertise in such areas as mechanics, mathematics, computer science, and robotics with an interest in biology, oceanography, and ecology. Most recently, she has been organizing groups of robotic underwater vehicles to create an adaptive sensor network. Leonard's work, characterized by an aptitude to delve into areas of study apart from her own engineering expertise, reflects her ability to collaborate with scientists from other fields, explore complex problems, and do sophisticated work that offers remarkable potential for applications outside its immediate arena. Leonard showed an early aptitude to cross-pollinate her ideas with different approaches while she was working on her PhD, which she received in 1994 at the University of Maryland, College Park. Under the direction of P.S. Krishnaprasad, Leonard was working on geometric control , particularly in underactuated robotic systems with restricted freedom of motion. Thinking about problems like controlling spacecraft orientation, Leonard developed theory that would help her systematically control motion in underactuated vehicles, a problem similar to parallel-parking a car. Making use of periodic control signals, averaging theory, and the geometry of (not necessarily linear) configuration spaces to predict how motion sequences in actuated directions would lead to motion in unactuated directions, she developed an algorithm for generating motions in any direction. Leonard asked Maryland professor David Akin if she could try her control algorithm by testing a vehicle in his Space Systems lab, a large neutral-buoyancy tank, in an experiment that contributed to her interest in the control of underwater vehicles. At Princeton, she began thinking about mechanics of bodies in water and considering ways to stabilize them. She based her approach on natural control forces such as gravity and used tools from geometric mechanics to determine the conditions for using these control forces to stabilize the vehicle's motion. " The idea was to think about using controls that could be realized by mechanical means so that the controlled system would still look like a mechanical system, " Leonard says. " The objective then was to choose among such controllers so that the controlled mechanical system was provably stable. " Around this time, Leonard met Jerry Marsden, now Carl F. Braun Professor of Engineering and Control of Dynamical Systems at the California Institute of Technology. Together with Anthony Bloch, professor of mathematics at the University of …