Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

ART: Adaptive Relational Transformer for Pedestrian Trajectory Prediction with Temporal-Aware Relations

Ruochen Li, Ziyi Chang, Junyan Hu, Jiannan Li, Amir Atapour-Abarghouei, Hubert P. H. Shum

Abstract

Accurate prediction of real-world pedestrian trajectories is crucial for a wide range of robot-related applications. Recent approaches typically adopt graph-based or transformer-based frameworks to model interactions. Despite their effectiveness, these methods either introduce unnecessary computational overhead or struggle to represent the diverse and time-varying characteristics of human interactions. In this work, we present an Adaptive Relational Transformer (ART), which introduces a Temporal-Aware Relation Graph (TARG) to explicitly capture the evolution of pairwise interactions and an Adaptive Interaction Pruning (AIP) mechanism to reduce redundant computations efficiently. Extensive evaluations on ETH/UCY and NBA benchmarks show that ART delivers state-of-the-art accuracy with high computational efficiency.

ART: Adaptive Relational Transformer for Pedestrian Trajectory Prediction with Temporal-Aware Relations

Abstract

Accurate prediction of real-world pedestrian trajectories is crucial for a wide range of robot-related applications. Recent approaches typically adopt graph-based or transformer-based frameworks to model interactions. Despite their effectiveness, these methods either introduce unnecessary computational overhead or struggle to represent the diverse and time-varying characteristics of human interactions. In this work, we present an Adaptive Relational Transformer (ART), which introduces a Temporal-Aware Relation Graph (TARG) to explicitly capture the evolution of pairwise interactions and an Adaptive Interaction Pruning (AIP) mechanism to reduce redundant computations efficiently. Extensive evaluations on ETH/UCY and NBA benchmarks show that ART delivers state-of-the-art accuracy with high computational efficiency.

Paper Structure

This paper contains 17 sections, 15 equations, 6 figures, 4 tables.

Table of Contents

  1. INTRODUCTION
  2. METHOD
  3. Problem Formulation
  4. Temporal-Aware Relation Graph Construction
  5. Adaptive Interaction Pruning
  6. Relational Transformer and Trajectory Decoding
  7. EXPERIMENTS
  8. Datasets and Evaluation Metrics
  9. Quantitative results
  10. Ablation Study and Model Analysis
  11. Ablation Study on Relation Weighting Strategies
  12. Ablation Study on Top-p Threshold
  13. Complexity and Efficiency Analysis
  14. Qualitative Results
  15. Trajectory Prediction Visualizations
  16. ...and 2 more sections

Figures (6)

  • Figure A1: Framework overview. Framework overview. The relation between pedestrians is inferred from the temporal evolution of their pairwise interactions over the observed history.
  • Figure A2: Overview of ART. Left: Temporal-Aware Relation Graph (TARG) leverages pairwise attention to model agent interactions across time steps, assigning higher weights to informative moments. Right: Adaptive Interaction Pruning (AIP) uses top-p filtering to adaptively retain informative neighbors based on cumulative interaction strength, producing a sparsified graph for trajectory prediction.
  • Figure C1: Ablation study of Top-$p$ threshold on the ETH/UCY dataset.
  • Figure C2: Qualitative comparisons with MART lee2024mart on the ETH/UCY dataset. Past trajectories are shown in blue, ground truth in red, and model predictions in green.
  • Figure C3: Qualitative comparisons with MART lee2024mart on the NBA dataset. Past trajectories are shown in blue, ground truth in red, and predictions in green.
  • ...and 1 more figures