UCSD ECE276B: Planning & Learning in Robotics (Spring 2022)

Time and Location

• Lectures: Tuesday and Thursday, 5:00 pm - 6:20 pm, in CENTR 214.

• Final Exam: Thursday, June 09, 2022, 7:00 pm - 10:00 pm in CENTR 105.

• Office Hours:

  • Tuesday, 6:20 pm - 7:00 pm, in CENTR 214

  • Thursday, 6:20 pm - 7:00 pm, in CENTR 214

  • Additional office hours: May 24, May 26, June 1, June 7, June 8, 5:00 pm - 6:00 pm on Zoom

Instructors

• Nikolay Atanasov: natanasov@eng.ucsd.edu

• Zhichao Li: zhl355@ucsd.edu

• Hanwen Cao: h1cao@ucsd.edu

Overview

This course covers optimal control fundamentals and their application to motion planning and decision making in robotics. Topics include Markov decision processes (MDPs), dynamic programming, search-based and sampling-based motion planning, value and policy iteration, linear quadratic regulation (LQR), and some model-free reinforcement learning.

Prerequisites

Students are expected to have programming experience at the level of a Data Structures and Algorithms course as well as background in statistical estimation or linear system theory at the level of at least one of the following UCSD courses:

• ECE276A: Sensing & Estimation in Robotics

• ECE271A: Statistical Learning I

• ECE275A: Parameter Estimation

• ECE225A: Probability and Statistics for Data Science

• ECE272A: Stochastic Processes in Dynamic Systems

• CSE250A: Principles of Artificial Intelligence: Probabilistic Reasoning and Learning

• CSE254: Statistical Learning

• MAE280A: Linear Systems Theory

Requirements

The class assignments consist of theoretical homework, a final exam, and three projects, each including a programming assignment in Python and a project report:

1. Dynamic Programming

2. Motion Planning

3. Infinite-horizon Optimal Control

Students are expected to sign up on Piazza and GradeScope:KY2BGP. Discussion and important announcements will be made on Piazza. The homework should be turned in and will be graded on GradeScope:KY2BGP.

Grading

Grading will be based on the following rubric.

References

• Dynamic Programming and Optimal Control: Bertsekas

• Reinforcement Learning: Sutton & Barto

• Planning Algorithms: LaValle

• Calculus of Variations and Optimal Control Theory: Liberzon

Collaboration and Academic Integrity

Please note that an important element of academic integrity is fully and correctly acknowledging any materials taken from the work of others. You are encouraged to work with other students and to discuss the assignments in general terms (e.g., “Do you understand the A* algorithm?” or “What is the update equation for Value Iteration?”). However, the work you turn in should be your own – you should not split parts of the assignments with other students and you should certainly not copy other students’ code or papers. All projects in this course are individual assignments. More generally, please familiarize yourself with UCSD's Code of Academic Integrity, which applies to this course. Instances of academic dishonesty will be referred to the Office of Student Conduct for adjudication.

IDEA Engineering Student Center

Please consider participating in the programs and events organized by the IDEA Engineering Student Center. The IDEA center, located to the right of the lobby of Jacobs Hall, is a hub for student engagement, academic enrichment, personal and professional development, leadership, community involvement, and a respectful learning environment for all. The IDEA center's mission is to foster an inclusive and welcoming community, promote academic success, develop engineering leaders, and, most importantly, support your mental health and wellness needs. These opportunities can be found on the IDEA Center Facebook page and the Center's web site.