Recursive State Estimation for Dynamic Systems
To present mathematical approaches to the best possible way of estimating the state of a general nonlinear dynamic system recursively, and to provide the implementation towards discrete-time systems in software based on typical applications in the field of object tracking and robotics.
After successful completion of this module, the students are able to give an overview of typical filtering approaches in a general discrete-time system, to explain the principles of different Gaussian, Bayesian and particle filters, to apply different filter approaches to data sets in the field of object tracking and robotic, to analyse application problems with regard to adequate system and observation models and to interpret predicted and filtered states obtained from the aforementioned filters correctly.
Module Contents
- optimal recursive state estimation in discrete-time systems (Kalman filter)
- Gaussian filters (extended Kalman filter, unscented Kalman filter and ensamble Kalman Filter) for nonlinear systems
- introduction into Bayesian inference
- the Bayes filter
- introduction into Monte Carlo techniques
- the particle filter
- applications to a tracking problems (e.g., regarding the motion of robots)
Lecturer
30167 Hannover
Exercise
30167 Hannover