Oliver Kroemer (Carnegie Mellon University), “Modularity and learning to structure robot manipulation skills”

Oliver Kroemer’s research focuses on developing algorithms and representations that enable robots to learn versatile manipulation skills over time. By equipping robots with the ability to acquire new skills and adapt manipulations to novel situations, his work opens up a wide range of potential applications—from assisting the elderly and maintaining parks and public spaces to operating in hazardous environments.

Oliver has developed methods that allow robots to learn about objects through physical interactions and autonomously refine their skills using reinforcement learning. Additionally, he has proposed innovative representations for capturing key aspects of manipulations, such as contact states and motor primitives, to enhance generalization across different tasks and scenarios.

The ultimate aim of his research is to create a life-long learning framework that enables robots to continuously acquire and improve manipulation skills, paving the way for more adaptable and capable robotic systems.

This talk will take place as part of SCIoI member Svetlana Levit’s seminar “Selected Topics in Robot Learning,” which explores how advances in machine learning are helping robots operate in new environments, learn new behaviors, and adapt to changing conditions.

Image generated with DALL-E by Maria Ott