Powered by OpenAIRE graph
Found an issue? Give us feedback
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/ NTNU Openarrow_drop_down
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
NTNU Open
Master thesis . 2024
Data sources: NTNU Open
addClaim

CART-SLAM: Cuda-Accelerated Real-Time SLAM

Authors: Tanberg, Eirik Lorgen;

CART-SLAM: Cuda-Accelerated Real-Time SLAM

Abstract

Det meste av forskning innen Simultaneous Localization And Mapping (SLAM) fokuserer på nøyaktigheten av metodene som mål på ytelse, og ser relativt lite på hvor raskt løsningene kjører. De begrenser seg også ofte til kun CPU-baserte implementasjoner. Det er mange gode grunner til dette. Det gjør utvikling enklere, og til nå har tilgjengeligheten av General-Purpose computing on Graphics Processing Units (GPGPU) maskinvare, spesielt i autonome integrerte systemer, vært begrenset. Med økende tilgjengelighet av denne typen maskinvare er mulighetene for å bruke disse til å forbedre ytelse tilstede. Dette prosjektet sentreres rundt segmentering av overflater i stereobilder for å oppgave hindringer med bruksområder innen autonom navigasjon ved å benytte Compute Unified Device Architecture (CUDA) for å forbedre ytelse. Vi presenterer flere nye metoder og fremgangsmåter for segmentering av overflater og superpikselsegmentering, inkludert en sanntidsimplementasjon av contour relaxation for superpikselsegmentering. Et rammeverk for behandling av data med høy parallellisme gjennomgående er også presentert. Systemet i seg selv, og modulene som er presentert i prosjektet, er gitt som åpen kildekode, og er laget med tanke på gjenbruk og videre arbeid. Resultatene viser at de foreslåtte fremgangsmåtene kjører i sanntid i både vei- og vann-scener, og gir lovende resultater innen segmentering av overflater i stereobilder, selv i utfordrende scenarioer som raske bevegelser og komplekse miljøer.

Most Simultaneous Localization And Mapping (SLAM) research focuses on the accuracy of the solution as the performance metric, and usually constrains their software to CPU only. There are many good reasons for this approach. It makes the development simpler, and until now the availability of General-Purpose computing on Graphics Processing Units (GPGPU) devices, especially in autonomous embedded systems, has been limited. With the general availability of GPGPU devices increasing, using these to improve performance is more feasible. This project centers around segmenting planes in stereo image pairs, motivated by detecting obstacles in autonomous navigation application, using Compute Unified Device Architecture (CUDA) to improve performance. We present several novelties for plane segmentation and superpixel segmentation, including a real-time implementation of contour relaxation for superpixel segmentation. A system framework to implement highly parallel processing systems is also presented. The system and its modules are provided as open-source software, and built with re-usability and expandability in mind. Results show that the proposed approaches run in real-time in automotive and maritime environment sequences, and yield promising results in segmentation planes in stereo images pairs, even in challenging scenarios with quick movement and complex scenes.

Country
Norway
  • BIP!
    Impact byBIP!
    selected citations
    These citations are derived from selected sources.
    This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    0
    popularity
    This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
    Average
    influence
    This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    Average
    impulse
    This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
    Average
Powered by OpenAIRE graph
Found an issue? Give us feedback
selected citations
These citations are derived from selected sources.
This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Citations provided by BIP!
popularity
This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
BIP!Popularity provided by BIP!
influence
This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Influence provided by BIP!
impulse
This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
BIP!Impulse provided by BIP!
0
Average
Average
Average
Green