Semantics in Robot Perception and Decision-Making Seminar Course

Wissenschaftliches Seminar

This course offers an introduction to semantics in robot perception and decision-making. In this course, students will work in pairs and conduct a literature review in the pertinent topic area of semantics in robot perception and decision-making. The topic areas include safe robot learning, learning from demonstration, language-conditioned robot learning, spatial AI, and 3D scene understanding; specific papers to be reviewed will vary for each term.

Learning Objectives

Students will learn to conduct a thorough literature review, contrast different ideas and concepts, quickly prototype existing algorithms, identify their impact and limitations, and communicate their findings in written and oral form. Furthermore, students will be able to structure their findings by identifying the main research threads in a particular field and engaging in group discussions and collaborative writing. 

Prerequisites

The prerequisites for this course are fundamental knowledge of robotics, machine learning, control theory, and computer vision.

Teaching and Learning Approach

Students will be given introductory lectures and guided through the process of finding relevant literature and discussing it. Regular meetings with a scientific advisor will support this process. In particular, the advisor will provide an introduction to the topic, an initial set of relevant literature, and early feedback on the report and presentation.

Evaluation

The evaluation for this course is composed of a written report (50%, length of report: up to 2 double-column pages per team member) and a presentation followed by questions and a discussion (50%, length of presentation: 5 minutes per team member).