Date: April 22, 2026
Speaker: Toru Lin, Ph.D .student, Berkeley AI Research (BAIR) Lab
Title: Learning Dexterous Robot Skills that Generalize: Embodied Intelligence from Autonomous Experience
Abstract: Robots hold the promise of assisting us in everyday environments, yet today’s systems remain limited in both dexterity (handling complex, contact-rich manipulation) and generalization (operating reliably in unstructured, changing scenes). The dominant approach in robot learning attempts to address this by scaling up datasets collected from human demonstrations. In this talk, I will argue that this approach faces fundamental limitations, and instead advocate for a shift toward learning from autonomous experience – where robots actively explore, interact with the environment, and improve through their own trial and feedback. I will outline the key building blocks required to make this paradigm practical on real robots, and present my research contributions to each. Together, these components enable robots to acquire dexterous skills that generalize across objects, tasks, and environments.
Bio: Toru Lin is a Ph.D. student at Berkeley AI Research (BAIR) Lab and a visiting researcher at Meta FAIR, advised by Jitendra Malik. Her research lies at the intersection of robotics and artificial intelligence, with a focus on learning dexterous and generalizable robot skills from autonomous experience. She received her BSc and MEng from MIT EECS, where she worked with Phillip Isola and Antonio Torralba, and previously studied at the University of Tokyo. She has interned at NVIDIA (GEAR group), DeepMind, Facebook, and Google. Her research is supported by the NSF Graduate Research Fellowship and the Berkeley Chancellor’s Fellowship.