Why the center of gravity in AI is moving from systems that talk to systems that see, understand, decide, and act. Robot learning comes first; hardware, control, and mechanics return as a late special module.
A personal companion page. The schedule below covers the full term; slides and recordings are for the sessions I teach, host, or co-run. Official announcements, assignments, deadlines, and grades live on the DGIST LMS.
Sessions
Aug 28Special lecture on AIOpening session, arranged by DGIST.DGIST
Sep 4The Paradigm Shift Toward Embodied AIFoundation models conquered language and images. Why action became the next frontier, and the map for everything that follows.LectureSlidesRecording
Sep 11Invited talkVideo Foundation Models and World ModelsHyeongmin Lee, Assistant Professor, SeoulTechDelivered remotely.SlidesRecording
Sep 18Leadership talkArranged by DGIST; speaker announced on the LMS.DGIST
Sep 25No class — Chuseok holidayNo class
Oct 2Action as a New Modality and the Scaling of Robot Learning DataAction treated as a data modality, and why robot data does not grow on its own the way web text and images did.LectureSlidesRecording
Oct 9No class — Hangul DayNo class
Oct 16Midterm week — a reflection report takes the place of an examNo sitting exam for this course. The prompt and the deadline are announced on the LMS.Assessment
Oct 23Methods for Learning Action: Imitation, Reinforcement Learning, and Diffusion PoliciesHow imitation and reinforcement learning differ, and how diffusion — already familiar from image generation — is reused to generate action.LectureSlidesRecording
Oct 30Leadership talk on entrepreneurshipArranged by DGIST.DGIST
Nov 6Invited talkSpatial AI to Physical AISunghwan Hong, Post-doc researcher, ETH ZurichDelivered remotely.SlidesRecording
Nov 13Leadership talkArranged by DGIST; speaker announced on the LMS.DGIST
Nov 20Special moduleFoundations of Robot Hardware, Control, and Modern PlatformsHow hardware, control, and mechanics turn a learned policy into physical motion.LectureSlidesRecording
Nov 27Ethics, Safety, and Social Responsibility in the Age of Embodied AIWhat changes once an AI system has a body and acts in the physical world: the new risk surface, labor, alignment, and privacy.LectureSlidesRecording
Dec 4Closing keynoteThe Physical AI Megatrend and the Coming Industrial RevolutionChris (Hongseop) Choi, CEO, Maum AIAn industry view of the market, the jobs, and the corporate strategy that acting AI is about to reshape. Delivered by my co-instructor on this course.SlidesRecording
Dec 11Final week — a synthesis essay takes the place of an examNo sitting exam for this course. The prompt and the deadline are announced on the LMS.Assessment
What this page covers
The Seminar for Comprehensive Competency Cultivation (SCCC, course code HSS118) is a required general-education course for first- and second-year students at DGIST, co-taught across a full sixteen-week term. This fall it is taught as Physical AI for Everyone. The schedule above lists every week so the shape of the term is visible at a glance. The eight highlighted sessions are the ones I lecture, host, or co-run; the weeks marked DGIST are arranged by the school, and both assessment weeks replace a sitting exam with written work.
Sessions assume no prior background in robotics or machine learning. Technical terms are introduced in plain language first, and every acronym is spelled out the first time it appears.
Slides and recordings
Links appear next to each session as material becomes available. Slides are published as web decks; recordings are released selectively. Material for invited sessions is published only where the speaker has agreed to it.
Contact
Course questions belong on the LMS so that every student sees the answer. For anything else, reach me at sung@maum.ai.