Unified Pedestrian Path Prediction Using Inverse Reinforcement Learning

2026-08-16

AI Summary

This study addresses the lack of unified learning frameworks and performance bottlenecks in complex models for pedestrian trajectory prediction by proposing a unified reinforcement learning paradigm based on Spatio-Temporal Graph Attention Networks (STGAT). By defining specialized state-action spaces and reformulating the decision-making process, this framework accommodates inverse reinforcement learning and diverse policy optimization algorithms, enabling universal improvements in graph model training objectives. Experimental results demonstrate that the proposed method significantly outperforms standard supervised learning across multiple benchmark datasets. These findings validate the effectiveness of task reformulation in enhancing advanced trajectory prediction architectures and offer novel perspectives for complex spatio-temporal modeling.

Abstract

Pedestrian path prediction is crucial for enhancing the safety of autonomous vehicles and advanced driver-assistance systems. Previous studies explored different learning-task formulations for pedestrian path prediction and compared these formulations using shallow neural networks, but did not extend this analysis to more complex deep-learning models. This paper adapts the Spatial-Temporal Graph Attention Network (STGAT) to a unified pedestrian path prediction framework and introduces state and action definitions specific to STGAT. The resulting formulations support deterministic and stochastic policies, one-time and sequential decision-making, and reinforcement-learning algorithms including REINFORCE and proximal policy optimization. The proposed learning-task formulations improve prediction performance across the selected benchmark datasets compared with the standard supervised-learning formulation. These results demonstrate that reformulating the decision process and training objective can improve an advanced pedestrian trajectory prediction architecture and may provide a path toward improving other graph-based prediction models.

Problem

• Pedestrian path prediction
• Learning-task formulations
• Deep learning models
• Inverse reinforcement learning

Innovation

• Inverse Reinforcement Learning
• Spatial-Temporal Graph Attention Network
• Unified Prediction Framework
• Pedestrian Path Prediction
• Decision Process Reformulation