Date: Wednesday, September 17, 2025
Speaker: Yuchen Cui, Assistant Professor of Computer Science, UCLA
Title: Interactive Robot Learning from Non-Expert Teachers
Abstract: Today’s general-purpose robot learning policies achieve only 50–80% zero-shot performance on downstream tasks. Bridging this gap requires robots that can continually adapt under human guidance. After deployment, however, most robots will interact not with trained experts but with everyday users. These non-expert teachers provide demonstrations and feedback that are noisy, inconsistent, and largely implicit—conditions that challenge traditional learning pipelines designed around clean, near-optimal data. My research addresses this gap by enabling robots to extract structure and meaning from natural, imperfect human interactions. The talk will explore two complementary solutions. First, I present a data quality framework for interpreting noisy human demonstrations. This includes a hybrid action representation to capture diverse modes of motion, a systematic study of how different types of demonstration modality affect learning, and a smoothness-driven metric for identifying higher-quality data. Second, I highlight how natural human signals, though noisy, can be leveraged as supervision. Here, I introduce a data-driven framework for treating facial reactions as implicit rewards, and show how large multimodal language models can be used to inject commonsense knowledge into the interpretation of gestures and gaze for grounding and task specification. Together, these approaches move toward robots that not only tolerate imperfect input, but actively benefit from the rich and varied signals that non-expert human partners naturally provide—enabling robots that can learn alongside people in everyday environments.
Bio: Yuchen Cui is currently an Assistant Professor of Computer Science at UCLA, directing the UCLA Robot Intelligence Lab. Prior to UCLA, she was a postdoc in the Computer Science Department at Stanford University and a fellow of the Stanford Institute for Human-Centered AI. Yuchen's research focuses on interactive robot learning and specifically on how to enable low-effort teaching for non-expert users. Yuchen obtained her Ph.D. in Computer Science from the University of Texas at Austin. Her dissertation is titled "Efficient algorithms for low-effort human teaching of robots”. During her graduate studies, Yuchen also conducted internships at Honda Research Institute, Diligent Robotics, and Facebook AI Research.
