Teaching & Theses

Type: Lecture

Semester: Winter Semester

Target Group: Master

This course will provide the theoretically and practically fundamentals of mobile robotics. The taught concepts and methods are key to an increasingly broad range of robot applications in domains such as households, intralogistics, service, industry, aerial vehicles, self-driving car.

Contents

1. Introduction
   1. Content of this Course
   2. Why Robotics?
2. Probability and Linear Algebra Refresher
   1. Laws of Probability, PGM, Bayesian Networks
   2. Matrices, Transforms
3. Robot Locomotion
   1. Kinematic Models
   2. Wheel Types, Sensors and Constraints
   3. Non-Holomonic Constraints
   4. Full Kinematic Analysis for Wheeled Robots
4. Uncertainty Propagation
   1. First-Order
   2. Monte Carlo
   3. Unscented Transform
5. Odometric Localization
   1. Integration Methods
   2. Differential Drive
   3. Three-Wheel Drive
   4. Other Odometry Models
6. Recursive Bayes Filter
   1. Temporal Reasoning and the State Space Model
   2. Inference Cases
   3. Recursive Bayes Filter
   4. Hidden Markov Models
7. Robot Localization
   1. Kalman Filter Localization
   2. Markov Localization
   3. Monte Carlo Localization
8. Maximal Likelihood Estimation
   1. Gauss-Newton Algorithm
   2. ICP and Odometry Calibration
9. Simultaneous Localization and Mapping
   1. EKF SLAM
   2. Graph SLAM
   3. RANSAC and Loop Closing
10. Grid Mapping
    1. Maximum Likelihood
    2. Maximum a Posteriori
11. Motion Planning
    1. Search
    2. Configuration Space
    3. Combinatorial Methods
    4. Sampling-Based Methods: PRM, RRT, RRT*

Lectures

• Hours: See Campus
• Room: See Campus
• Language: English

Exercises

• We reinforce each chapter with an exercise, including comprehension questions, paper-and-pencil tasks, and programming tasks. Assignments will be published after the lecture and have to be submitted the following week before the exercise session
• Hours: See Campus. In general, we have 90 minute session in which we discuss the solution of the previous exercise and brief you on the next one
• Room: See Campus
• Language: English

Exam

• Form: Written (120 min)
• Admission requirements: 70% exercises solved or sensibly dealt with, and submitted

FAQ

• Admission to course? No limit, but please un-register if you are not taking it or change your mind 
• Requirements? The course has no formal requirements (e.g. other lectures, classes), we assume basics in probability, statistics, linear algebra, system theory, programming
• Are both needed, classes and exercises? Yes! We will drop students that are enrolled in only one
• Recordings? No recordings! We want you to go out and interact
• Slides? Will be made available on a weekly basis on ILIAS. Design paradigm: in between a full script and bullet-point slides
• Exam discussion? Yes, typically in the last lecture we present style and concept of the written exam
• Erasmus? We love Erasmus and other exchange programs, but we cannot provide special treatments, e.g. individual remote exams, individual exam requirements, etc. Organize yourself that you can be physically present for the exam.

Type: Lecture

Semester: Summer Semester

Target Group: Master

Requirements: Mobile Robotics

The lecture covers advanced theoretical foundations and practical techniques for intelligent mobile robots. It is intended as a continuation of our introductory lecture "Mobile Robotics," which students should have previously attended. The content includes both classical and learning-based methods for problems such as visual SLAM, place recognition, motion planning or 3D environment modeling. We will also give an introduction to Human-Robot Interaction and spotlights on current research topics in robotics.

For the practical part, we will provide introductions to C++ and ROS and give out two programming projects.

Content
1.  Robot Programming: Introduction to C++ and ROS 
2.  Visual Odometry and SLAM
3.  Motion Planning
4.  Human-Robot Interaction
5.  Robot Learning Examples: Robot Reinforcement Learning, Deep Learning and Foundation Models for Robotics
6.  Resarch Spotlights

• Exercises: two programming projects
• Number of students: max. 20
• Exam: oral
• Language: English

Type: Fachpraktikum / practical course

Semester: Summer Semester

Target Group: Master

Robots navigating their environments rely on exteroceptive sensors such as cameras (stereo, monocular, RGB-D) and LiDARs to perceive and model the world around them. In this course, the focus is on the use of cameras for robot navigation. Key topics include the fundamentals of visual perception systems in robotics, processing image data streams, and transitioning between 3D environments and 2D projections. The course emphasizes building a visual odometry pipeline in C++/Python, with opportunities to extend the work by integrating proprioceptive sensors (e.g., IMUs), generating 3D maps with tools like nvBlox or Neural Radiance Fields, adding semantic information, or developing full visual SLAM systems. Work is based on both publicly available datasets and real-world data collected with camera systems.

Type: Advanced Seminar (Hauptseminar)

Semester: Every Semester

Target Group: Master

The seminar introduces students to state-of-the-art research in robotics and teaches them how to efficiently and critically engage with scientific literature and the paper’s code. Example areas include 3D computer vision, 3D scene understanding, human behavior modeling and transfer, imitation learning, dexterous mobile manipulation, and social human-robot interaction, as well as the underlying techniques from classical robotics, deep and reinforcement learning, or foundation models.

Seminar Structure

The seminar consists of two parts. In the first part, each student will select one recent scientific paper from a curated list. He/she will read, understand, analyze, and present the paper. In the second part, the students will choose from two options:

1. Literature track (LT): Conduct an in-depth literature survey on the paper’s topic, e.g. to trace back the roots of the idea
2. Code track (CT): Run, reproduce or extend the paper’s implementation (Note that we cannot provide compute infrastructure from our side, the students need to have the means to run the code).

Tasks and Deliverables

1. Understand and analyze the selected paper to answer key research questions (see below).
2. Give two presentations (approx. 15 minutes + 5 minutes discussion each), one focusing on the paper itself, and one focusing on the literature analysis (LT) or experiments (CT)
3. Write a survey report (we will teach you how to do this) for the LT or submit your code with a concise report on your experiment setup for the CT.
4. Actively participate in seminar sessions, including discussions of all presented papers.

The final grade will be based on presentation, report or implementation, and active participation.

 Key Questions (LT and/or CT)

• What is the research question addressed in the paper? (LT, CT)
• Why is it relevant to the field? (LT, CT)
• How does it relate to other work? (LT)
• What are the paper’s contributions and hypotheses? (LT, CT)
• How are experiments designed to support the hypotheses? (CT)
• Can the results be reproduced or extended under different conditions? (CT)
• What are the strengths, weaknesses, and limitations of the approach?  (LT, CT)
• What scientific ideas inspired the work? (LT, CT)
• How could you take the work further?  (LT, CT)

Learning Goals

• Critically reading and analyzing scientific literature (in robotics and related fields) 
• Understanding experimental design and scientific methodology.
• Presenting technical content clearly and engagingly.
• Running and interpreting robotics code (if applicable).
• Engaging in scholarly discussions on advanced research topics.  

Information

• Organizer: Prof. Kai Arras
• Co-Organizers: Dennis Rotondi, Till Hielscher, Fabio Scaparro, Tim Nickel, Abdelrhman Werby
• Language: English
• The seminar is limited to 12 students, first come first serve

We are excited to invite you to participate in the Robotics Oberseminar, a weekly forum for Master's and PhD students. (This is an internal event, registration via Campus is not possible).

Our mission is to create a supportive and informal space to bridge the multi-disciplinary worlds of robotics, computer vision, machine learning, or human-robot interaction. It's a great opportunity to connect with peers, get constructive feedback on your work, and stay up-to-date on the latest research across our fields.

What we do in our sessions:

• Discuss influential papers: From foundational works to state-of-the-art research.
• Share research progress: Get feedback on your thesis or PhD project in a low-pressure environment.
• Learn new skills: Participate in hands-on tutorials (e.g., Docker, JAX, ROS 2).
• Practice for conferences: Rehearse your talks and posters with a friendly audience.
• Hear conference takeaways: Get summaries and highlights from major conferences (CVPR, ICRA, etc.).
• When: Wednesday at 11:30-13:00, Where: U 32.122

Type: Studienprojekte B.Sc. SWT/MI oder Entwicklungsprojekte M.Sc. SWT

Semester: Every Semester

Target Group: Bachelor and Master

Autonomous mobile robots are being put to practical use in a growing number of applications: as self-driving cars, household aids, transportation platforms in intralogistics, or last-mile delivery drones in urban areas. From a software engineering perspective, autonomous robots represent an exciting challenge: a multitude of complex algorithms must be integrated with hardware connections and under real-time requirements as functional blocks in a multi-layered architecture.

In our Studien- or Entwicklungsprojekte, a team of 5 to 9 students develops, integrates, tests and deploys a fully autonomous robotic system. While tasks vary, recent projects focused on developing the SnackBot, a delivery robot for office environments. SnackBot autonomously fulfills snack delivery orders within the AI Institute’s U32 building. Orders are placed and tracked using a dedicated web-based GUI, allowing every institute member to easily order their favorite snack delivered directly to their desk.

We simulate the conditions of real agile software projects with a role-play and a fictitious budget: the customer has a limited time and resource budget and only a vague idea of the product. Infrastructure and support generate costs, while successful deliveries generate revenue. In a fictitious cost accounting system, the goal is to achieve profitability and user satisfaction.

Interested? Enrollment of these projects are organized by Katrin Schneider, Program Manager and Erasmus Coordinator of the Computer Science Department. Hint: organize yourself and form a team already prior to the assignment process.