International Scholar Spotlight
An Interview with International Scholar Michael Yip
(Q): A summary of your research - understandable to those outside your field of expertise.
(A): My lab research is in surgical robotics and robot learning. We take both a physical prototyping approach to solving these problems, as well as a computational approach. For surgical robotics, we design new minimally invasive, endoscopic or snake-like robotic instruments and systems to perform diagnostic and therapeutic tasks deep within the body – far deeper than conventional tools would be able to access. These robotic tools enable enhanced dexterity far beyond current capabilities and offer a platform for autonomous surgical tasks: robot-enhanced biopsy, foreign bodies removal, ablation therapy, and tumor resection. These tasks, with robot control, could be done under real-time medical image-guidance at sub-millimeter precisions which go beyond human capabilities.
Computationally, we are interested in how a robot learns to control themselves in an effective way --- i.e. what is the architecture for learning robust and useful behaviors --- which involve topics in deep reinforcement learning and robot planning. Much of this has to do with reinforcement learning, which involves a robotic system learning to behave through interactions with the environment and collecting experience. Our interest stems in finding data-efficient reinforcement learning for safety-critical tasks: i.e. robots should learn quickly (so a library of tasks can be learned efficiently), safely (to avoid wear and tear and to avoid damaging themselves or the environment), and generally (so that learned features can be transferred to new tasks).
While my lab, the Advanced Robotics and Controls Lab, was established less than three years ago, our research has led to the development of millimeter diameter flexible robotic catheters and endoscopes that can navigate deep within the human body and through conventional scopes (such as colonoscopes, bronchoscopes, rhinolaryngoscopes) and provide significant dexterity beyond human controlled tools. We have furthermore developed fundamental algorithms for rapid robot learning and exploration for planning behaviors in complex environments, leading to multiple orders of magnitude increase in computational and power efficiency. We even starting a new project with the San Diego Zoo to use robots to help them take care of their animals.
(Q): Why did you choose UC San Diego?
(A): UC San Diego offered the strongest intersection of opportunities in machine learning, robotics, and medicine. Indeed, with the world-class UCSD Medical School, the Contextual Robotics Institute, and the Halıcıoğlu Data Science Institutes, as well as the surrounding community of high-tech biomedical startups and military contractors, the opportunities for robotics and medical devices development puts it within the top three places in the U.S. to do this type of research. Finally, with support and interest in robotics in general across all levels of governance (i.e. Chancellor’s Office to student body), the decision was easy. There is also a hunger for engagement between local industry – who wish to build San Diego into a U.S. robotics hub that I only appreciated when I arrived, and I am excited about the opportunities to work with the many local companies, large and small, in making San Diego the premier robotics destination.
(Q): An anecdote in transitioning to life in San Diego.
(A): Life in San Diego is a joy and transitioning here has been relatively painless. There is a diverse set of cultures in the area, unmatched outdoor life, close proximities to most amenities, and fantastic and diverse food options. The only challenge is to make sure that you have the time outside of work to enjoy all that San Diego has to offer!
