Back to projects

Robust Control

  • Graduate Student: Martin Fevre
  • Principal Investigator: Dr. James Schmiedeler & Dr. Bill Goodwine
  • Team: Martin Fevre & Tan Chen
  • Funding: National Science Foundation Robust Intelligence Program
  • Goal: To design and implement robust controllers on 5-link underactuated biped to reject velocity disturbances and yield energetically efficient walking consistent with human locomotion.

Legged robots must be both robust to disturbances and energetically efficient. While underactuated bipeds offer great energetic efficiency similar to human locomotion, their robustness mainly depends on their ability to reject disturbances in the unactuated degrees of freedom (DOF). Control laws have been developed to stabilize robots following a velocity disturbance but applications are too often limited to ankle-actuated flatfooted robots for which robustness outweighs efficiency.

Velocity decomposition of underactuated biped robots, however, quantifies the nonlinear coupling between the links and the control authority over the unactuated DOF. This measure of control authority can be used to design controllers that make use of the amount of nonlinear coupling available to yield robust and efficient walking.