Extended target tracking using Gaussian processes on stick-pixel defined objects
In this work, I present the performance of a extended target tracking algorithm that utilizes Gaussianprocesses. The extended target tracking algorithm is evaluated on objects corresponding to roadusers, with automotive use in mind. The measurements that denes the object to be tracked arederived fro...
Main Author: | |
---|---|
Format: | Others |
Language: | English |
Published: |
Mälardalens högskola, Akademin för innovation, design och teknik
2017
|
Subjects: | |
Online Access: | http://urn.kb.se/resolve?urn=urn:nbn:se:mdh:diva-35782 |
id |
ndltd-UPSALLA1-oai-DiVA.org-mdh-35782 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-UPSALLA1-oai-DiVA.org-mdh-357822017-06-17T05:32:01ZExtended target tracking using Gaussian processes on stick-pixel defined objectsengOlsson, AndersMälardalens högskola, Akademin för innovation, design och teknik2017RoboticsRobotteknik och automationIn this work, I present the performance of a extended target tracking algorithm that utilizes Gaussianprocesses. The extended target tracking algorithm is evaluated on objects corresponding to roadusers, with automotive use in mind. The measurements that denes the object to be tracked arederived from stereo image sensor disparity and is called Stick-pixels or Stixels. The process ofgenerating these measurements are also presented in this work.It consists of two separate methods,one relying on stereo image frames and one purely dened by object characteristics and pose. Theextended target tracking algorithm has been tested on three types of simulated road users, car, cyclistand pedestrian. To evaluate the performance of the target tracking algorithm three measures areused, error in position, orientation discrepancy over time and intersection over union. Student thesisinfo:eu-repo/semantics/bachelorThesistexthttp://urn.kb.se/resolve?urn=urn:nbn:se:mdh:diva-35782application/pdfinfo:eu-repo/semantics/openAccess |
collection |
NDLTD |
language |
English |
format |
Others
|
sources |
NDLTD |
topic |
Robotics Robotteknik och automation |
spellingShingle |
Robotics Robotteknik och automation Olsson, Anders Extended target tracking using Gaussian processes on stick-pixel defined objects |
description |
In this work, I present the performance of a extended target tracking algorithm that utilizes Gaussianprocesses. The extended target tracking algorithm is evaluated on objects corresponding to roadusers, with automotive use in mind. The measurements that denes the object to be tracked arederived from stereo image sensor disparity and is called Stick-pixels or Stixels. The process ofgenerating these measurements are also presented in this work.It consists of two separate methods,one relying on stereo image frames and one purely dened by object characteristics and pose. Theextended target tracking algorithm has been tested on three types of simulated road users, car, cyclistand pedestrian. To evaluate the performance of the target tracking algorithm three measures areused, error in position, orientation discrepancy over time and intersection over union. |
author |
Olsson, Anders |
author_facet |
Olsson, Anders |
author_sort |
Olsson, Anders |
title |
Extended target tracking using Gaussian processes on stick-pixel defined objects |
title_short |
Extended target tracking using Gaussian processes on stick-pixel defined objects |
title_full |
Extended target tracking using Gaussian processes on stick-pixel defined objects |
title_fullStr |
Extended target tracking using Gaussian processes on stick-pixel defined objects |
title_full_unstemmed |
Extended target tracking using Gaussian processes on stick-pixel defined objects |
title_sort |
extended target tracking using gaussian processes on stick-pixel defined objects |
publisher |
Mälardalens högskola, Akademin för innovation, design och teknik |
publishDate |
2017 |
url |
http://urn.kb.se/resolve?urn=urn:nbn:se:mdh:diva-35782 |
work_keys_str_mv |
AT olssonanders extendedtargettrackingusinggaussianprocessesonstickpixeldefinedobjects |
_version_ |
1718460562132697088 |