Vehicle Trajectory Prediction based on LSTM Recurrent Neural Networks

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

3 Citations (Scopus)
29 Downloads (Pure)

Abstract

This work presents an effective tool to predict the future trajectories of vehicles when its current and previous locations are known. We propose a Long Short-Term Memory (LSTM) Recurrent Neural Network (RNN) prediction scheme due to its adequacy to learn from sequential data. To fully learn the vehicles' mobility patterns, during the training process we use a dataset that contains real traces of 442 taxis running in the city of Porto, Portugal, during a full year. From experimental results, we observe that the prediction process is improved when more information about prior vehicle mobility is available. Moreover, the computation time is evaluated for a distinct number of prior locations considered in the prediction process. The results exhibit a prediction performance higher than 89%, showing the effectiveness of the proposed LSTM network.

Original languageEnglish
Title of host publication2021 IEEE 93rd Vehicular Technology Conference, VTC 2021-Spring - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781728189642
DOIs
Publication statusPublished - Apr 2021
Event93rd IEEE Vehicular Technology Conference, VTC 2021-Spring - Virtual, Online
Duration: 25 Apr 202128 Apr 2021

Publication series

NameIEEE Vehicular Technology Conference
PublisherInstitute of Electrical and Electronics Engineers Inc.
Volume2021-April
ISSN (Print)1550-2252

Conference

Conference93rd IEEE Vehicular Technology Conference, VTC 2021-Spring
CityVirtual, Online
Period25/04/2128/04/21

Keywords

  • Deep Learning
  • Long Short-Term Memory (LSTM) Network
  • Recurrent Neural Networks (RNNs)
  • Trajectory Prediction
  • Transportation Data Analytics

Fingerprint

Dive into the research topics of 'Vehicle Trajectory Prediction based on LSTM Recurrent Neural Networks'. Together they form a unique fingerprint.

Cite this