Hyper-STTN: Hypergraph Augmented Spatial-Temporal Transformer Network for Trajectory Prediction

TL;DR

Hyper-STTN introduces a hypergraph-augmented spatial-temporal transformer that effectively models complex group and pairwise interactions for accurate crowd trajectory prediction, outperforming existing methods.

Contribution

The paper presents a novel hypergraph-based spatial-temporal transformer network that captures multiscale groupwise and pairwise interactions for improved trajectory prediction.

Findings

Hyper-STTN outperforms state-of-the-art baselines on public datasets.

Multiscale hypergraph modeling enhances group interaction representation.

Fusion of heterogeneous features improves prediction accuracy.

Abstract

Predicting crowd intentions and trajectories is critical for a range of real-world applications, involving social robotics and autonomous driving. Accurately modeling such behavior remains challenging due to the complexity of pairwise spatial-temporal interactions and the heterogeneous influence of groupwise dynamics. To address these challenges, we propose Hyper-STTN, a Hypergraph-based Spatial-Temporal Transformer Network for crowd trajectory prediction. Hyper-STTN constructs multiscale hypergraphs of varying group sizes to model groupwise correlations, captured through spectral hypergraph convolution based on random-walk probabilities. In parallel, a spatial-temporal transformer is employed to learn pedestrians' pairwise latent interactions across spatial and temporal dimensions. These heterogeneous groupwise and pairwise features are subsequently fused and aligned via a multimodal transformer. Extensive experiments on public pedestrian motion datasets demonstrate that Hyper-STTN consistently outperforms state-of-the-art baselines and ablation models.