CMSC 818B: Decision-Making for Robotics - Fall 2019

• Instructor: Pratap Tokekar (lastname at umd.edu)
• Lectures: Mon & Wed, 2:00 PM - 3:15 PM in Atlantic Building 1114

Class announcements, discussions, assignments, and due dates will be posted on TBD.

Description

This course will focus on the principles of decision-making in robotics. The motion planning course in Spring '17 focused on the problem of how to go from point A to point B. In this course, we will examine higher-level algorithms that make decisions as to what points A and B should be (for example) depending on the task at hand. Typical tasks that we will use as prototypical examples are where robots as mobile sensors. Specific research topics include:

• Combinatorial Vehicle Routing: Traveling Salesperson Problem, Orienteering, Distributed Voronoi Coverage, Multi-Robot Assignment
• Learning and Perception: Gaussian processes, Reinforcement Learning, Vision-Based Planning
• Information-theoretic path planning: Adaptive sampling, Submodular Optimization, Multi-armed bandit problems, Optimal stopping problems
• Planning under Uncertainty: Chance-constrained optimization, Risk-aware planning, Resilient Coordination

Most of the lectures will be based on foundational papers. This will be followed by an in-depth study of recent research papers which will be led by the students. The students will also work on a research-oriented course project. The goal of the course is to understand and critique research papers, identify open problems, and initiate progress towards solving them. An ideal outcome is for students to write a research paper at the end of the course.

The students will also work in groups on a research-oriented course project. More information about the project is here.

This is a qualifying course in artificial intelligence.

Grading

This is still tentative. Evaluation will be composed of the following:

• Homework (40%)
• Exams (25%)
• Project (25%)
• Class participation + presentation (10%)

Textbook

No required textbook. Course materials will be drawn from research papers and notes that I will post online.

Schedule

Assignments

Details will posted on canvas.

Paper Presentations

Each student will lead the discussion on TBD papers during the semester. You may choose a paper from the list given here.

Course Project

The project can be any of the following types:

• Survey: An in-depth survey of a research area. The deliverable is a survey paper (between 8-10 pages).
• Theoretical: Work on formulating a new research problem and/or solving an existing problem. It is perfectly okay if you cannot solve the problem in its entirety during the class, however, you must demonstrate sufficient effort and progress. The deliverable is a project report (between 8-10 pages).
• Implementation: Implement an existing or new algorithm. This can be entirely in software or on hardware. The deliverable is a demo in the last week of classes and a report (6 pages).

All reports must follow the IEEE Transactions format.

Prerequisites

This is a seminar-style course that will feature a discussion of research papers. An ability to read and critique research papers is required.

There are no explicit prerequisites. However, students should be very comfortable in at least two of the following areas: Probability and Bayesian Statistics; Algorithms; Machine Learning/AI; Optimization.

Prior background in robotics is helpful, but not required.

Academic Integrity

You are encouraged to discuss the course materials with the instructor and other students in the class. However, any work that you submit (including but not limited to homeworks, paper reviews, project reports) must be your own. Give proper citations if you use any code or data from anyone else.

Services for Students with Disabilities

Any student who feels that he or she may need an accommodation because of a disability (learning disability, attention deficit disorder, psychological, physical, etc.), please make an appointment to see me during office hours.