Computational Design of Hardware and Software for Autonomous Robots
February 16, 2018 - 11:15am to 12:15pm
Carnegie Mellon University
3-180 Keller Hall
Biography: Sehoon Ha is currently a postdoctoral fellow at Carnegie Mellon University (CMU). Before joining CMU, he worked at Disney Research Pittsburgh as an associate research scientist. He received his Ph.D. degree in Computer Science from the Georgia Institute of Technology. Sehoon’s research lies at the intersection between computer graphics and robotics, including physics-based control, fabrication, design optimization, and deep reinforcement learning. He developed computational approaches to automate the complex robot development process by jointly optimizing hardware and software. He also proposed a set of general computational methods for designing a variety of agile motion controllers for virtual and real robots.
Abstract: Developing hardware and software for a new robot is traditionally a complex process that requires the time-consuming efforts of experienced engineers. The development of an automated and computational design pipeline supported by solid mathematical predictions would help researchers to more readily create new highly capable and cost effective robots. However, this paradigm of computer-aided robot design introduces unique challenges including those relating to interrelating different layers of the robot development process, including mechanical design, optimal control, and behavior learning. My ongoing work tackles these challenges by building mathematical models and developing interactive software combining computational powers with human intuition. In the long term, my goal is to develop the computational methods that underpin a rich ecosystem of robotic devices that have the potential to improve various aspects of our lives, including new search-and-rescue robots that effectively negotiate the cluttered environment and custom medical surgery robots that are tailored to individual patients.