LG-Traj: LLM Guided Pedestrian Trajectory Prediction

TL;DR

LG-Traj leverages Large Language Models and advanced trajectory clustering techniques within a transformer framework to significantly improve pedestrian trajectory prediction accuracy in dynamic environments.

Contribution

This paper introduces LG-Traj, a novel method that integrates LLM-generated motion cues and trajectory augmentation to enhance pedestrian trajectory prediction.

Findings

Improved prediction accuracy on ETH-UCY and SDD benchmarks.

Effective use of LLMs for generating motion cues.

Enhanced trajectory representations through SVD augmentation.

Abstract

Accurate pedestrian trajectory prediction is crucial for various applications, and it requires a deep understanding of pedestrian motion patterns in dynamic environments. However, existing pedestrian trajectory prediction methods still need more exploration to fully leverage these motion patterns. This paper investigates the possibilities of using Large Language Models (LLMs) to improve pedestrian trajectory prediction tasks by inducing motion cues. We introduce LG-Traj, a novel approach incorporating LLMs to generate motion cues present in pedestrian past/observed trajectories. Our approach also incorporates motion cues present in pedestrian future trajectories by clustering future trajectories of training data using a mixture of Gaussians. These motion cues, along with pedestrian coordinates, facilitate a better understanding of the underlying representation. Furthermore, we utilize singular value decomposition to augment the observed trajectories, incorporating them into the model learning process to further enhance representation learning. Our method employs a transformer-based architecture comprising a motion encoder to model motion patterns and a social decoder to capture social interactions among pedestrians. We demonstrate the effectiveness of our approach on popular pedestrian trajectory prediction benchmarks, namely ETH-UCY and SDD, and present various ablation experiments to validate our approach.