Simulating Realistic Local 3D Maps from Geometric Models
Simulace realistickĂ˝ch lokálnĂch 3D map z geometrickĂ©ho modelu
Authors
Supervisors
Reviewers
Editors
Other contributors
Journal Title
Journal ISSN
Volume Title
Publisher
České vysoké učenà technické v Praze
Czech Technical University in Prague
Czech Technical University in Prague
Date of defense
Files
Abstract
TerĂ©nnĂ robot by se mÄ›l dokázat samostatnÄ› pohybovat pĹ™es obtĂĹľnĂ˝ terĂ©n. UÄŤenĂ jĂzdy pĹ™es pĹ™ekážky probĂhá v simulovanĂ©m prostĹ™edĂ. SimulovanĂ˝ senzor, kterĂ˝ pĹ™edává robotovi informace o terĂ©nu, má vĹľdy o okolĂ perfektnĂ informace, na rozdĂl od reálnĂ©ho senzoru RealSense na robotu, ve kterĂ©m se vyskytuje šum a chybÄ›jĂcĂ data. CĂlem práce je vytvoĹ™it metodu, která dokáže zkreslit simulátorová data tak, aby vypadala realisticky. VycházĂme z našà pĹ™edchozĂ práce, která vyuĹľĂvá CyklickĂ© sĂtÄ› GAN k vytvářenĂ realistickĂ˝ch dat pro jinĂ˝ senzor. LepšĂch vĂ˝sledkĹŻ dosahujeme Ăşpravou parametrĹŻ, pĹ™edzpracovánĂm vstupnĂch dat, a pĹ™idánĂm párovanĂ˝ch dat.
An outdoor terrain robot should be able to traverse difficult terrain autonomously. This is learned in a simulated environment to make the process faster and easier. However, the simulated sensor used for terrain mapping in this environment has always precise information, which is very different from reality where digital elevation model (DEM) is noisy and incomplete. The goal is to develop a method that would process DEM created by simulation and transform it to a more realistic DEM. We build on top of our previous work that uses Cycle GAN networks to generate realistic DEMs from another sensor. We improve current results by adjusting parameters, preprocessing inputs, and introducing paired data.
An outdoor terrain robot should be able to traverse difficult terrain autonomously. This is learned in a simulated environment to make the process faster and easier. However, the simulated sensor used for terrain mapping in this environment has always precise information, which is very different from reality where digital elevation model (DEM) is noisy and incomplete. The goal is to develop a method that would process DEM created by simulation and transform it to a more realistic DEM. We build on top of our previous work that uses Cycle GAN networks to generate realistic DEMs from another sensor. We improve current results by adjusting parameters, preprocessing inputs, and introducing paired data.
Description
Citation
Underlying research data set URL
Permanent link
Rights/License
A university thesis is a work protected by the Copyright Act of the Czech Republic. Extracts, copies and transcripts of the thesis are allowed for personal use only and at one`s own expense. The use of thesis should be in compliance with the Copyright Act.
Vysokoškolská závÄ›reÄŤná práce je dĂlo chránÄ›nĂ© autorskĂ˝m zákonem. Je moĹľnĂ© poĹ™izovat z nÄ›j na svĂ© náklady a pro svoji osobnĂ potĹ™ebu vĂ˝pisy, opisy a rozmnoĹľeniny. Jeho vyuĹľitĂ musĂ bĂ˝t v souladu s autorskĂ˝m zákonem v platnĂ©m znÄ›nĂ.
Vysokoškolská závÄ›reÄŤná práce je dĂlo chránÄ›nĂ© autorskĂ˝m zákonem. Je moĹľnĂ© poĹ™izovat z nÄ›j na svĂ© náklady a pro svoji osobnĂ potĹ™ebu vĂ˝pisy, opisy a rozmnoĹľeniny. Jeho vyuĹľitĂ musĂ bĂ˝t v souladu s autorskĂ˝m zákonem v platnĂ©m znÄ›nĂ.