Research in our lab seeks to discover, describe, and utilize the core principles of control of animal movement. This research program is rooted in the belief that nature has found fundamentally different approaches for getting physical tasks done, compared to the usual methods of robotics engineers. In particular, we focus on legged locomotion of multi-legged animals, and on producing robust and maneuverable motions from unreliable bodies in poorly characterized environments.
The research program consists of three main parts:
- Theoretical development of mathematical ideas that could be used for modeling animal and robot locomotion. This consists of work in the fields of dynamical systems theory and control theory.
- Data-driven characterization of legged locomotion: the development of computational tools that allow mathematical descriptions of locomotion to be quantitatively identified and produced from observational data taken from animals or machines. Work here consists of robotic science and collaboration with biologists.
- Development of inexpensive, modular, rapidly evolving robot architectures. These allow us to face the a similar design problem to that faced by Nature: getting an imprecisely manufactured body, with slow communication and a modular structure, to reliably perform locomotion tasks in a poorly modeled environment.