Tracking of Partially Visible Elliptical Objects with a Lidar Sensor using Random Matrices and a Virtual Measurement Model
- Virtual measurement models (VMM) can be used to generate artificial measurements and emulate complex sensor models such as Lidar. The input of the VMM is an estimation and the output is the set of measurements this estimation would cause. A Kalman filter with extension estimation based on random matrices is used to filter mean and covariance of the real measurements. If these match the mean and covariance of the artificial measurements, then the given estimation is appropriate. The optimal input of the VMM is found using an adaptation algorithm. In this paper, the VMM approach is expanded for multi-extended object tracking where objects can be occluded and are only partially visible. The occlusion can be compensated if the extension estimation is performed for all objects together. The VMM now receives as input an estimation for the multi-object state and the output are the measurements that this multi-object state would cause.
Author: | Patrick HoherORCiD, Johannes ReuterORCiD, Felix GovaersORCiD, Wolfgang Koch |
---|---|
DOI: | https://doi.org/10.1109/SDF55338.2022.9931696 |
ISBN: | 978-1-6654-8672-9 |
ISBN: | 978-1-6654-8673-6 |
Parent Title (English): | Sensor Data Fusion: Trends, Solutions, Applications (SDF), 12-14 Oct. 2022, Bonn, Germany |
Publisher: | IEEE |
Document Type: | Conference Proceeding |
Language: | English |
Year of Publication: | 2022 |
Release Date: | 2022/11/14 |
Tag: | Multi-extended object tracking; Random matrices; Lidar; Virtual measurement model; Extension estimation; Occlusion |
Page Number: | 6 |
Note: | Volltextzugriff für Angehörige der Hochschule Konstanz via Datenbank IEEE Xplore möglich |
Institutes: | Institut für Systemdynamik - ISD |
Relevance: | Keine peer reviewed Publikation (Wissenschaftlicher Artikel und Aufsatz, Proceeding, Artikel in Tagungsband) |
Open Access?: | Nein |
Licence (German): | Urheberrechtlich geschützt |