ROB 498 /599: Computational Human-Robot Interaction Fall 2024

Instructor: Christoforos Mavrogiannis (cmavro at umich dot edu)

GSI: Jeeho Ahn (jeeho at umich dot edu)

Lectures: Mon/Wed 15:00 - 16:30 (3150 DOW)

Class material: Piazza; Canvas

Office hours: Christoforos: Mon/Wed, 16:30-17:30 (FRB 3310); Jeeho: Wed, 2:00-3:00 (FRB 3171)

Course description: This is a special-topics course (3 units), covering computational techniques that enable robots to work with and around people. Topics include estimation, planning, and control algorithms, discussed in the context of applications like crowd navigation, collaborative manipulation, and assistive robotics. Besides algorithmic foundations, the course will explore topics in experiment design, discussing evaluation methodologies that will enable smooth deployment of robots in human environments. Through student-led paper presentations and a team project, students will gain exposure to the state of the art in computational HRI.

Learning objectives: In this class, students will gain exposure to computational techniques used to develop human-robot interaction (HRI) applications and systems and get familiar with the process of interpreting and presenting research. Specifically, by the end of the class, students will be able to:

• Recognize and understand the components of an HRI system.
• Understand the importance of accounting for users when developing algorithmic frameworks for HRI.
• Understand the landscape of essential computational tools for developing HRI applications.
• Design a user study to evaluate an HRI system.
• Interpret and critically analyze the methodology of research papers in HRI (and beyond).
• Communicate effectively research methodologies and evidence to a peer audience.

Prerequisites: There are no formal prerequisites but mathematical maturity (e.g., ROB 101, Math 215, IOE 265) and programming background (e.g., ROB 320 or EECS 281) are expected. A foundation on the design of human-robot systems (e.g., ROB 204) is recommended.

Textbook: There is no official textbook. Background for most of the course components can be found in the book Computational Human-Robot Interaction by Thomaz, Hoffman and Cakmak (pdf). Background on probability and filtering can be found in Probabilistic Robotics by Thrun, Burgard, and Fox (pdf). Additional background on planning can be found in Planning Algorithms by Lavalle (pdf).

Grading:

Grading will be based on the performance along the following components:

• Paper Presentations (35%): Students will present assigned papers in class, on a rotating schedule. Evaluation will be based on demonstrated effort in understanding the technical content, the depth of the discussion and its relationship to the class themes, the clarity, structure and timing of the presentation, and the ability to respond to questions.
• Project (40%): Students will develop and work on team projects to solidify their understanding of Computational HRI and get practical experience. Students are expected to devote between 60-80 hours over two months (October, November) to the project. Project teams will be graded based on their comprehensive presence throughout the semester across the deliverables and specifically based on the demonstrated mastery of the literature and the technical content, the novelty of the contribution the insight in the categorization of previous work, the quality of the final presentation.
• Quizzes (15%): At the beginning of every lecture, there will be a short (10 minute) quiz on the material from the previous lecture and the readings for the day. This quiz will not be challenging and is only meant to motivate a consistent contact with the material throughout the semester.
• Participation (10%): This class is intended to be interactive, and discussion driven. Thus, all students are expected to ask questions and participate in the discussions. Every class will feature a dedicated time for questions and discussion.

Syllabus: A more detailed syllabus document can be found here.

Acknowledgements: This class is inspired by HRI classes at USC, Cornell, and Berkeley.