Traffic-Aware Pedestrian Intention Prediction

2025-07-16

AI Summary

Existing pedestrian intention prediction models neglect dynamic traffic signals and scene context, hindering safe decision-making for autonomous vehicles (AVs). To address this, we propose the Traffic-Aware Spatio-Temporal Graph Convolutional Network (TA-STGCN), the first method to jointly model traffic light states (red/yellow/green) and pedestrian bounding box dimensions as node attributes in a graph structure. This explicitly encodes the coupling between signal phase and pedestrian spatial scale, thereby enhancing modeling of complex spatio-temporal dependencies in urban environments. Built upon graph convolutional networks, TA-STGCN fuses multi-source spatio-temporal features—including traffic signal timing, pedestrian motion trajectories, and contextual scene cues. Evaluated on the PIE dataset, it achieves a 4.75% absolute improvement in prediction accuracy over strong baselines and significantly outperforms current state-of-the-art methods. Our approach establishes a novel paradigm for safety-critical, interpretable, and traffic-aware pedestrian intention prediction.

Technology Category

• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding

Abstract

Accurate pedestrian intention estimation is crucial for the safe navigation of autonomous vehicles (AVs) and hence attracts a lot of research attention. However, current models often fail to adequately consider dynamic traffic signals and contextual scene information, which are critical for real-world applications. This paper presents a Traffic-Aware Spatio-Temporal Graph Convolutional Network (TA-STGCN) that integrates traffic signs and their states (Red, Yellow, Green) into pedestrian intention prediction. Our approach introduces the integration of dynamic traffic signal states and bounding box size as key features, allowing the model to capture both spatial and temporal dependencies in complex urban environments. The model surpasses existing methods in accuracy. Specifically, TA-STGCN achieves a 4.75% higher accuracy compared to the baseline model on the PIE dataset, demonstrating its effectiveness in improving pedestrian intention prediction.

Problem

• Predict pedestrian intention using traffic signals and context
• Improve accuracy in autonomous vehicle navigation safety
• Integrate dynamic traffic states into spatio-temporal modeling

Innovation

• Integrates traffic signs and states into prediction
• Uses dynamic signal states and bounding box features
• Employs spatio-temporal graph convolutional network