Inhalt des Dokuments
Teaching:Inertial Sensor Fusion
From Fachgebiet Regelungssysteme TU Berlin
Contents |
Announcements
This course will be offered as an online course in WS20/21. If you are interested in attending the course, please enroll for the corresponding ISIS course. The password is "a poetry style in which the theme is divided into four parts".
General Information
- Course Number
- 0430 L 097
- Institution
- 0430 Institut für Energie- und Automatisierungstechnik
FG Regelungssysteme - Type
- 4 IV
- Cycle
- weekly
- Allowability
- 6 CP Master, Bachelor upon application to examination office
- Teachers
- Dr.-Ing. Thomas Seel (Lectures) and MSc Dustin Lehmann (Computer Exercises)
- Teaching Language
- English
- Exam
- Portfolio exam with practical tasks (documented source code) and written task (multiple-choice tests)
Content
Participants will acquire knowledge on inertial sensor technologies for motion tracking of aerial/ground/water vehicles, robotic actuators and human body segments. They will learn and apply methods for sensor fusion and analysis of inertial mesurement data. This includes methods to estimate the orientation and position of moving objects in three-dimensional space as well as methods for calculating joint angles or classifying human motion. Since most of the considered applications are feedback control problems, the course focuses on realtime algorithms. The methods will be applied to application data during designated computer lab instructions integrated in this module.
Topics include, but are not limited to:
- Basic principles of gyroscopes, accelerometers and magnetometers
- Error characteristics of MEMS-based inertial sensors
- Application: Gait phase detection by foot-worn inertial sensors
- Quaternions and other representations of 3D rotations
- Orientation estimation from inertial measurement data
- Application: Position tracking/retrieval of an unmanned aerial vehicle
- Joint angle estimation from inertial measurement data
- Application: Realtime motion tracking of a robotic actuator
- Kalman filtering methods for linear and nonlinear systems
- Probabilistic sensor fusion and Bayesian state estimation
- Identification of kinematic parameters from inertial measurement data
- Application: Human body motion tracking by wearable inertial sensors
Schedule
The course takes place weekly throughout the regular teaching period of the semester. All teaching material will be available online for asynchronous studying. Every week, a synchronous online meeting of about 90 minutes will be dedicated to answering questions and discussing the taught concepts. The time slot for these meetings is open for optimization based on the teachers' and the participants' availabilities.
Exam
Portfolio exam consisting of homework tasks (eight small assignments, 2% each), one practical task (source code and extended abstract on a given research problem, 14%) and written tasks (two short exams, each 35%). Grading is based on Notenschlüssel 3 (40%=4.0, 45%=3.7, ..., 80%=1.3, 85%=1.0). The tentative(!) dates for the short written exams are December 14 and February 22. I always encourage students to create a poll for optimization of exam dates. If you find a date that suits everybody better than the current dates, please let me know.
Downloads
All downloads will be made available via a course website on the Moodle-based eLearning platform of the university. The link and password are provided above in the announcements.