Data Analytics for Cyber-Physical Systems: Automatic Failure Diagnosis

Type Praktikum
Semester Winter Semester 2020/21
Language English

Ehsan Zibaei

Course Details Preliminary Meeting Slides


As robotic systems are increasingly employed in different applications, more and more safety incidents occur during their operation. One way to improve the safety of robotic systems is to look into the vast amount of data produced by them and automatically learn the causes of the incidents. This way the next versions of the robotic system will be designed safer against similar faults. Causal discovery algorithms are powerful tools that use statistical dependency between variables to infer the causal structure and hence reveal the causes of the system failure. The focus of this practical course is on using causal inference methods to discover the causal relationship between events in the context of robotic systems in general and Unmanned Aerial Vehicles as a specific use case. The objectives of this practical course include:

- Perform safety analysis for robotic systems
- Learn causality concepts and causal discovery methods
- Compare causal discovery algorithms based on their characteristics


The course sessions will take place every week online with continuous report submissions. The course has three phases: First the dataset of events and system failures will be preprocessed. Then causal models will be inferred from the dataset. Finally the inferred causal relationships will be verified by experimentation in the simulation environment.


If you want to be prioritized by us in the matching system, please apply through our application form. Please apply until the 17th of July to be considered.

Rules for participation

  1. Plagiarism of any form (blatant copy-paste, summarizing someone else's ideas/results without reference etc.) will result in immediate expulsion from the course.
  2. All submissions are mandatory. Each submission must fulfill a certain level of quality. Submissions that are just collections of buzzword/keywords or coarse document structures will not be accepted. Failing that will be graded 5.0.
  3. Late submissions will invite penalties.
  4. Non-adherence to the submission guidelines will invite penalties.
  5. Participation and attendance in all practical course meetings is mandatory.


[1] Spirtes, Peter, and Kun Zhang. "Causal discovery and inference: concepts and recent methodological advances." Applied informatics. Vol. 3. No. 1. Springer Berlin Heidelberg, 2016.

[2] Center for Causal Discovery. 2020. Video Tutorials. [online] Available at:

[3] PX4. 2020. Open Source For Drones PX4 . [online] Available at: