kth.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Online learning for data-driven trajectory predictions
KTH, School of Electrical Engineering and Computer Science (EECS).
2024 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesisAlternative title
Onlineinlärning för datadrivna banförutsägelser (Swedish)
Abstract [en]

Accurate trajectory predictions of surrounding actors, such as vehicles and pedestrians, are crucial for ensuring the safety of autonomous driving systems and other road agents. The primary problem addressed in this thesis is the limited accuracy and robustness of existing trajectory prediction methods, particularly when dealing with anomalies and the need for continuous model updates. To tackle this problem, the thesis explores deep learning methods and online learning approaches. The main contribution is the development of an online learning algorithm that incorporates adaptive learning rate and edge case detection. This algorithm aims to improve trajectory prediction accuracy by dynamically adjusting the learning rate based on the distributional properties of the data. The results demonstrate the promising improvements achieved by the proposed online learning algorithm in trajectory prediction accuracy. The algorithm provides valuable insights into the significance of adaptive learning and the integration of anomaly detection in trajectory prediction for autonomous driving.

Abstract [sv]

Exakta prediktioner av trajektorier för omgivande aktörer, såsom fordon och fotgängare, är avgörande för att säkerställa säkerhet och effektivitet i system för autonom körning. Denna avhandling fokuserar på att undersöka och förbättra noggranhet och robusthet hos befintliga metoder för prediktering av trajektorier. Mer specifikt fokuserar detta arbete på att hantera avvikelser och variationer i datat genom att utforska djupinlärningsmetoder och algoritmer för kontinuelig inlärning. Denna avhandling fokuserar på att undersöka och förbättra noggranhet och robusthet hos befintliga metoder för prediktering av trajektorier. Mer specifikt fokuserar detta arbete på att hantera avvikelser och variationer i datat genom att utforska djupinlärningsmetoder och algoritmer för kontinuelig inlärning. Algoritmen uppviar lovande resultat vid kontinuelig inlärning med förbättrad noggranhet i trajektioieprediktionerna. Arbetet ger värdefulla insikter om betydelsen av kontinuelig och adaptiv inlärninggenom avvikelsedetektering i traktorieprediktering för autonom körning.

Place, publisher, year, edition, pages
2024. , p. 85
Series
TRITA-EECS-EX ; 2024:439
Keywords [en]
Trajectory prediction, Online learning, Autonomous driving
Keywords [sv]
Banförutsägelse, Onlineinlärning, Autonom körning
National Category
Computer and Information Sciences
Identifiers
URN: urn:nbn:se:kth:diva-351376OAI: oai:DiVA.org:kth-351376DiVA, id: diva2:1887424
External cooperation
Scania AB
Supervisors
Examiners
Available from: 2024-09-19 Created: 2024-08-07 Last updated: 2024-09-19Bibliographically approved

Open Access in DiVA

fulltext(6298 kB)353 downloads
File information
File name FULLTEXT01.pdfFile size 6298 kBChecksum SHA-512
dde61275d7e1ff5bc4ab4568869c45310ed5e0341e4f9eaa7b68bcf45ea2b2eb907c09f6222563ce18290fd662fbc6db797a94e944190e76153d5330575913aa
Type fulltextMimetype application/pdf

By organisation
School of Electrical Engineering and Computer Science (EECS)
Computer and Information Sciences

Search outside of DiVA

GoogleGoogle Scholar
Total: 353 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 199 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf