StudiumAbschlussarbeiten
Rao-Blackwellized Extended Kalman Particle Filter with Implicit and Explicit Measurement Equations for Georeferencing of Multi-Sensor-Systems (MSSs)

Rao-Blackwellized Extended Kalman Particle Filter with Implicit and Explicit Measurement Equations for Georeferencing of Multi-Sensor-Systems (MSSs)

© Moftizadeh
Betreuung:  Rozhin Moftizadeh, Hamza Alkhatib
E-Mail:  moftizadeh@gih.uni-hannover.de
Jahr:  2022

A Multi-Sensor-System (MSS) refers to various installed sensors on a single platform that are frequently used in engineering to capture different aspects of an environment. In order to combine the derived data for further analysis purposes, it is important for the MSS to be georeferenced. In other words, it is essential to derive the pose of the MSS with respect to a global coordinate system. In practical applications, the most straightforward way of georeferencing is to directly use the Global Navigation Satellite System (GNSS) data to determine the 3D position of the MSS. However, when it comes to urban environments, due to a number of destructive effects such as multipath and shadowing, such data might either not be available or reliable enough to be used. On the other hand, regardless of the coordinate system, the Inertial Measurement Unit (IMU) sensors can always deliver the 3D orientation of the MSS. However, due to drifting over time, there are always errors and biases in the IMU data, which makes their direct use critical. Therefore, other solutions should be sought to best deal with the MSS georeferencing task. One of these solutions is to use some additional information, which might be available in the surrounding environment of the MSS. Such additional information is usually available within 3D city maps. However, these maps are usually prone to generalization errors, which affects their reliability. Therefore, it is important to detect these errors and count for them while estimating the MSS pose by including these maps. Moreover, the framework that is used for the georeferencing task should be able to on the one hand encounter the global uncertainties, which are usually irresistible when it comes to real applications, and on the other hand it should be computationally efficient.

Consequently, the main aim of the current master thesis is to develop a particle filtering framework that while preserving efficiency, it can simultaneously deliver the MSS pose and a corrected map of the environment.