Research

Paper

TESTING March 09, 2026

Context-free Self-Conditioned GAN for Trajectory Forecasting

Authors

Tiago Rodrigues de Almeida, Eduardo Gutierrez Maestro, Oscar Martinez Mozos

Abstract

In this paper, we present a context-free unsupervised approach based on a self-conditioned GAN to learn different modes from 2D trajectories. Our intuition is that each mode indicates a different behavioral moving pattern in the discriminator's feature space. We apply this approach to the problem of trajectory forecasting. We present three different training settings based on self-conditioned GAN, which produce better forecasters. We test our method in two data sets: human motion and road agents. Experimental results show that our approach outperforms previous context-free methods in the least representative supervised labels while performing well in the remaining labels. In addition, our approach outperforms globally in human motion, while performing well in road agents.

Metadata

arXiv ID: 2603.08658
Provider: ARXIV
Primary Category: cs.LG
Published: 2026-03-09
Fetched: 2026-03-10 05:43

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.08658v1</id>\n    <title>Context-free Self-Conditioned GAN for Trajectory Forecasting</title>\n    <updated>2026-03-09T17:37:03Z</updated>\n    <link href='https://arxiv.org/abs/2603.08658v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.08658v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>In this paper, we present a context-free unsupervised approach based on a self-conditioned GAN to learn different modes from 2D trajectories. Our intuition is that each mode indicates a different behavioral moving pattern in the discriminator's feature space. We apply this approach to the problem of trajectory forecasting. We present three different training settings based on self-conditioned GAN, which produce better forecasters. We test our method in two data sets: human motion and road agents. Experimental results show that our approach outperforms previous context-free methods in the least representative supervised labels while performing well in the remaining labels. In addition, our approach outperforms globally in human motion, while performing well in road agents.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.LG'/>\n    <published>2026-03-09T17:37:03Z</published>\n    <arxiv:comment>Accepted at the 2022 21st IEEE International Conference on Machine Learning and Applications (ICMLA)</arxiv:comment>\n    <arxiv:primary_category term='cs.LG'/>\n    <author>\n      <name>Tiago Rodrigues de Almeida</name>\n    </author>\n    <author>\n      <name>Eduardo Gutierrez Maestro</name>\n    </author>\n    <author>\n      <name>Oscar Martinez Mozos</name>\n    </author>\n    <arxiv:doi>10.1109/ICMLA55696.2022.00196</arxiv:doi>\n    <link href='https://doi.org/10.1109/ICMLA55696.2022.00196' rel='related' title='doi'/>\n  </entry>"
}