David Inkyu Kim

University of Southern California, Embedded Systems (United States)

Papers

Total Citations

H-Index

About

David Inkyu Kim is a leading researcher in robotic perception and human-robot interaction, whose work bridges the gap between raw sensor data and actionable robot understanding. His core research focuses on **affordance-based reasoning**—the idea that robots should not just see objects and spaces, but understand what actions they enable (e.g., "pushable," "liftable," or "sleepable"). Kim’s seminal 2014 paper on semantic labeling of 3D point clouds using object affordances (66 citations) pioneered a technique that extracts geometric features to classify objects by their functional potential, directly enabling more intuitive robot manipulation. He advanced this concept with his work on interactive affordance map building (21 citations), where a robot learns spatial relationships through a Markov Random Field model, allowing it to dynamically understand its environment. In his 2016 study on human-centric spatial affordances (9 citations), Kim extended this framework to human activity recognition, demonstrating that semantic place labels like "kitchen" or "bedroom" can predict activities such as cooking or sleeping. By grounding robotic intelligence in functional, human-relevant categories, Kim’s research has laid critical groundwork for robots that can assist humans in unstructured, real-world settings.