Announcements
General Information
Times & Places
MoWeFr 12:00PM - 12:50PM, RWAC 0103
Course Staff
| Name | ||
|---|---|---|
| Instructor | Hao Su | haosu@ucsd.edu |
| TA | Yulin Liu | yul266@ucsd.edu |
Office Hours
- Instructor's Office Hour: For questions related to lectures, logistics, etc.
Fri 3pm-4pm, Zoom (to find the link, see Piazza annoucement) - TA's Office Hour: For general questions related to homeworks, grading, etc.
Thu 11am-12pm, CSE 240B
Overview
This course introduces the framework of Embodied AI, which aims to develop intelligent agents that can interact with the physical world. The course will cover key concepts of this emerging field with assignments and projects, including embodiment, reinforcement learning, imitation learning, 3D perception, real2sim, and sim2real technologies.
The course has both technical width and depth, and expect the students to have a strong background in calculus, linear algebra, and deep learning. There will be a significant amount of coding assignments.
Prerequisites
- Strong background in calculus and linear algebra.
- Project experience in deep learning.
- Familiar with Newtonian mechanics.
- Proficient with Python.
- Experience in physical simulation is a plus.
Grading (tentative)
The course includes mandatory homeworks and optional homeworks. The load of the course will be relatively heavy.Syllabus
The planned syllabus is as below. Certain contents may be added or removed based upon the interactions in class and other situations.
- Classical robotics pipeline (sense, planning and control)
- 2D and 3D computer vision
- Cognitive basis of Embodied AI
- Reinforcement learning and imitation learning
- Data strategies of Embodied AI
Acknowledgements
Thank Hillbot Inc. for support.