Control development for robotic systems has led to ever-increasing precision, speed, robustness, repeatability, and agility for industrial, aerial and ground robots. The relative novelty of magnetically levitated (maglev) robots presents fertile ground for controls research, and the complexity of high degree of freedom (DOF) humanoid robots operating on uncertain terrain has long challenged standard control techniques. The effects of real-world effects, such as wind, on the performance of vertical-take-off-and-landing (VTOL) flying robots are also relatively unknown. Further, changes in robot characteristics over time and environmental disturbances not seen in controlled settings leave many existing control techniques vulnerable to failure in the field.

Our goal with this research theme is to dramatically improve individual and multi-robot control performance. We propose using ‘learning controllers’ that allow robots to perform more precise motions over time and path following methods that determine when it is safe to progress toward a goal. As these challenges are addressed, coordination of multiple robots in the same environment will become possible, and new issues in coordinated control will arise, particularly for high DOF robots such as humanoids.

With a fixed motion capture system, an extensive collection of various types of robots, and a collection of interchangeable environment components, a wide range of control challenges can be examined in the RoboHub, enabling efficient and safe experimentation and accelerated development.

Within this research theme, we will focus primarily on four topics: