Abstract:
Dynamic interactions play a fundamental role in human capabilities, enabling us to achieve a wide range of tasks such as moving heavy objects, manipulating our surroundings, and changing directions rapidly and safely. In contrast, most conventional robotic systems lack this level of agility and cannot perform dynamic interactions, limiting their potential in practical applications.

In this talk, I will explore pushing off from a surface as an example of dynamic interactions to change the direction of motion. I will show that push-off skills can be captured by simple models that can easily be transferred from simulation to the real robot. I will further discuss how experts (us humans) utilize wall push-offs to quickly change their direction of motion. A similarly simple underlying structure can be found in the motions, and I will show that the pushing skill can be represented through a simple data-driven model based on human motion capture data and force measurements. Finally, I will introduce and demonstrate a transfer process that allows us to adapt and employ this skill model to different robot embodiments.

Committee:
Nancy Pollard
Oliver Kroemer
Zeynep Temel
Arpit Agarwal