Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MobilityDL: A Review of Deep Learning From Trajectory Data

Anita Graser, Anahid Jalali, Jasmin Lampert, Axel Weißenfeld, Krzysztof Janowicz

TL;DR

MobilityDL surveys deep learning approaches applied to trajectory data across eight mobility use cases, emphasizing a data-centric view that maps dense, sparse, and aggregated trajectories to DL architectures like RNNs, CNNs, Transformers, and GNNs. It catalogs data representations (sequences, grids, graphs, images), models, and evaluation metrics, and highlights the importance of data engineering in enabling DL on mobility data. The review reveals DL’s prominence in traffic-crowd prediction and trajectory-related tasks, but also notes limited benchmarking rigor, reproducibility challenges, and variable transferability across regions. Key contributions include a structured taxonomy by use case, architecture, and data granularity, plus trend analysis showing increasing attention and graph-based methods. The work underscores the need for benchmarks, explainability, and sustainable tooling to advance real-world adoption of DL for mobility trajectories.

Abstract

Trajectory data combines the complexities of time series, spatial data, and (sometimes irrational) movement behavior. As data availability and computing power have increased, so has the popularity of deep learning from trajectory data. This review paper provides the first comprehensive overview of deep learning approaches for trajectory data. We have identified eight specific mobility use cases which we analyze with regards to the deep learning models and the training data used. Besides a comprehensive quantitative review of the literature since 2018, the main contribution of our work is the data-centric analysis of recent work in this field, placing it along the mobility data continuum which ranges from detailed dense trajectories of individual movers (quasi-continuous tracking data), to sparse trajectories (such as check-in data), and aggregated trajectories (crowd information).

MobilityDL: A Review of Deep Learning From Trajectory Data

TL;DR

MobilityDL surveys deep learning approaches applied to trajectory data across eight mobility use cases, emphasizing a data-centric view that maps dense, sparse, and aggregated trajectories to DL architectures like RNNs, CNNs, Transformers, and GNNs. It catalogs data representations (sequences, grids, graphs, images), models, and evaluation metrics, and highlights the importance of data engineering in enabling DL on mobility data. The review reveals DL’s prominence in traffic-crowd prediction and trajectory-related tasks, but also notes limited benchmarking rigor, reproducibility challenges, and variable transferability across regions. Key contributions include a structured taxonomy by use case, architecture, and data granularity, plus trend analysis showing increasing attention and graph-based methods. The work underscores the need for benchmarks, explainability, and sustainable tooling to advance real-world adoption of DL for mobility trajectories.

Abstract

Trajectory data combines the complexities of time series, spatial data, and (sometimes irrational) movement behavior. As data availability and computing power have increased, so has the popularity of deep learning from trajectory data. This review paper provides the first comprehensive overview of deep learning approaches for trajectory data. We have identified eight specific mobility use cases which we analyze with regards to the deep learning models and the training data used. Besides a comprehensive quantitative review of the literature since 2018, the main contribution of our work is the data-centric analysis of recent work in this field, placing it along the mobility data continuum which ranges from detailed dense trajectories of individual movers (quasi-continuous tracking data), to sparse trajectories (such as check-in data), and aggregated trajectories (crowd information).
Paper Structure (16 sections, 6 figures, 7 tables)

This paper contains 16 sections, 6 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Representing Trajectory Data For Deep Learning Use Cases
  3. Trajectory prediction / imputation
  4. Arrival time prediction
  5. (Sub)trajectory classification
  6. Anomaly detection
  7. Next location / destination prediction
  8. Synthetic data generation
  9. Location classification
  10. Traffic volume prediction
  11. Why Deep Learning?
  12. DL Motivations
  13. Baselines & Metrics
  14. Discussion
  15. Trends
  16. ...and 1 more sections

Figures (6)

  • Figure 1: Overview of trajectory data representations used to train neural networks
  • Figure 2: Portions of publications by use case from 2018 to 2023. Total number of publications per year is displayed at the top of each bar.
  • Figure 3: Portions of publications by NN design from 2018 to 2023. Total number of publications per year is displayed at the top of each bar.
  • Figure 4: Overview of the deep learning approaches per use cases from 2016 to 2023
  • Figure 5: Publication counts by use case from 2018 to 2023.
  • ...and 1 more figures