Information-Based Georeferencing of Multi-Sensor-Systems by Particle Filter with Implicit Measurement Equations

EasyChair Preprint 6420

Abstract

Multi-Sensor-System (MSS) georeferencing is a challenging task in engineering that should be dealt with in the most reliable way possible. The most straight forward way for localizing a MSS is to rely on the Global Navigation Satellite System (GNSS) and Inertial Measurement Unit (IMU) data. However, these data might not be always reliable enough or even available. Therefore, suitable filtering techniques are required to deal with such problems and to increase the reliability of the estimated states. In global localization and when it comes to real scenarios, particle filters are proven to deliver more realistic results than Kalman filter realizations. However, in MSS georeferencing where multiple sensors are used, different observation models are needed some of which could be of implicit type. In such a case, the likelihood estimation is challenging due to impossibility of estimating the observations by means of the generated samples. Therefore, the current paper offers a new particle filter methodology that can handle both implicit and explicit observation models. Final results of this methodology, which is applied on a simulated environment for georeferencing a MSS, are shown to be satisfactory.

Keyphrases: 6-DOF, MSS, Monte Carlo simulation, georeferencing, implicit observation model, particle filter

@booklet{EasyChair:6420,
  author    = {Rozhin Moftizadeh and Sören Vogel and Alexander Dorndorf and Jan Jüngerink and Hamza Alkhatib},
  title     = {Information-Based Georeferencing of Multi-Sensor-Systems by Particle Filter with Implicit Measurement Equations},
  howpublished = {EasyChair Preprint 6420},
  year      = {EasyChair, 2021}}