Authors: Armin Danesh Pazho, Vinit Katariya, Ghazal Alinezhad Noghre, Hamed Tabkhi

Published on: November 11, 2023

Impact Score: 8.22

Arxiv code: Arxiv:2311.06623

Summary

• What is new: Introduction of a novel transformer-based approach, VT-Former, that combines long-range temporal pattern recognition with a new Graph Attentive Tokenization (GAT) module for understanding social interactions among vehicles.
• Why this is important: The need for enhanced roadway safety and traffic management through accurate vehicle trajectory prediction.
• What the research proposes: VT-Former, a new method leveraging transformer technology and GAT modules for superior vehicle trajectory prediction in various traffic and safety applications.
• Results: State-of-the-art performance on three benchmark datasets, demonstrating VT-Former’s generalizability, robustness, and efficiency on embedded systems, along with potential in vehicle anomaly detection.

Technical Details

Technological frameworks used: Transformer-based models

Models used: VT-Former, Graph Attentive Tokenization (GAT)

Data used: Three benchmark datasets with different viewpoints

Potential Impact

Intelligent transportation systems, traffic management solutions, roadway safety technology providers, energy conservation in transportation, autonomous vehicle development companies

Want to implement this idea in a business?

We have generated a startup concept here: TrajectoNet.