Inferring Missing Trajectory Data with Temporal Convolutional Networks
Abstract
Trajectory data collected in real-world settings is frequently incomplete due to sensor failure, communication loss, or occlusion. We address the task of \emph{trajectory inpainting}: reconstructing contiguous missing segments from observed context. We propose a Temporal Convolutional Network (TCN) with symmetric dilation that relaxes the standard causality constraint, allowing each time step to draw on both past and future observations, a property that is essential for inpainting, but absent from forecasting-oriented architectures. The model is trained with a composite loss that combines weighted mean squared error, boundary--continuity penalties, and a smoothness regularizer. Trained on a synthetic dataset of (train), (validation), and (test) two-dimensional trajectories with randomly placed 20% masked segments, the model achieves good R, MSE and MAE metrics.
Cite
@article{arxiv.2607.25147,
title = {Inferring Missing Trajectory Data with Temporal Convolutional Networks},
author = {Ilinca Tiriblecea and Gabriel Turinici},
journal= {arXiv preprint arXiv:2607.25147},
year = {2026}
}