Harold Park
Biography
Fun Fact: I've free soloed a 35m tall climb!
Hi! I’m Sejun and I was born in South Korea, moving to the U.S. at the age of 8. Since then I’ve moved frequently, finally ending up at the Bay Area where I’ve lived my last 4 years. I’ve always been a numbers person, and after being introduced to olympiad math, I’ve looked forward to being involved in the APMA-CS concentration at Brown. Some of my hobbies include water sports, classical music, and playing poker and golf!
Research
Over the past summer, I conducted research in the Humans to Robots lab at Brown, mentored by PhD candidate Benned Hedegaard. My project revolves around a planning question: when you give a robot a goal like "erase the whiteboard in the other room," it has to work out both a sequence of steps and the physical details that make each one possible. Reasoning around the raw detail such as the robot's joint angles is computationally impossible (because there is just too much information), so the robot needs a simplified symbolic picture of the world to plan over which is a categorical abstraction of these low level details. My work was building the representations of objects the robot reasons about and collecting the data that defines when each of its skills can run and what changes afterward. The symbolic picture alone isn't enough though, since a plan that looks fine at that level can turn out to be physically impossible once the robot tries it, like reaching for a handle it can't actually get to from where it's standing. So the two levels have to run together, with motion planning checking whether each step is achievable and the robot backtracking to a different plan when it isn't. I was initially very interested in projects within category theory, and through this project I have learned the difference between theoretical soundness versus the accuracy in physical implementation.