Course Description

Robot learning is a challenge at the heart of AI and robotics to enable robot to make complex, sequential decisions to act effectively in diverse environments. This graduate-level course combines lectures with paper readings. In this course, we will introduce core ideas from machine learning, deep learning, vision and language, behavior cloning, and how such approaches are applied to robot decision-making and control. Through lectures and research paper discussions, students will explore key methods such as behavior cloning, multimodal perception, and policy learning, while examining open challenges in embodied intelligence and gaining broad insights for your research.

Administrative Information

• Lectures: Mon/Wed, 10:10-11:25 AM
• Dates: 08/18/2025 -12/12/2025
• Instructor: Mingyu Ding
• Location: SN011
• Teaching Assistants: Yunchao Yao (yunchaoy@cs.unc.edu)
• Requirements: This is a graduate-level course that requires some mathematical and machine learning background.

Grading and Course Policies

• Grading:

  • Attendancy (20%)

  • Paper presentation (30%)

  • Final course project (50%)

• Collaboration: Collaboration on presentation and course projects are allowed.

• Late Submissions: All assignments are due on the respective due date at 11:59 pm Eastern Time. Only on-time assignments will be accepted.

• Academic Integrity: You are permitted to utilize external resources for assignments, provided that proper acknowledgment is given in the appropriate location. For more details about the honor code, see UNC Honor Code Description.

Forms

Paper Selection Form: https://forms.gle/i3qnzE4ZmMDjsi9JA

Schedule

• Lecture 1 (08/18): Introduction to Robot Learning

• Lecture 2 (08/20): Robot Perception

• Lecture 3 (08/25): 3D Vision, Transformation

• Lecture 4 (08/27): TBD