Data-Driven Fine Manipulation in Robotics - 2024 Winter Robotics Colloquium

Attend a compelling robotics colloquium featuring Liyiming Ke from the Allen School, focusing on data-driven fine manipulation. Explore innovative approaches to automating delicate tasks like cutting fingernails, threading needles, and performing intricate surgical procedures using general-purpose robotic hardware. Discover how Ke's research leverages imitation learning and reinforcement learning to develop precise, robust, and adaptive policies for fine manipulation challenges. Learn about frameworks that enhance the robustness of imitation learning agents and training paradigms that improve sample efficiency in reinforcement learning, potentially surpassing human capabilities with minimal data. Gain insights into combining learning methods with structures and priors to reduce human effort in automation while improving robotic precision and robustness. Hear from Ke, a final-year PhD candidate at the University of Washington and Rising Star in EECS 2023, about her work on pushing the boundaries of robotic fine motor skills, including the development of a low-cost chopsticks robot platform for fine manipulation in dynamic environments.