Social Graph Mamba: Forecasting Pedestrian Movements Based on Social Context
This work is significant for robotics and autonomous navigation systems that need to operate safely and efficiently in human-populated spaces, offering an incremental improvement in computational efficiency and group behavior modeling.
The paper addresses the challenge of forecasting pedestrian movements in crowded environments, which is crucial for autonomous navigation. It proposes Social Graph Mamba (SGM) to replace attention-based social reasoning with Selective State Space Models (SSMs) on dynamic interaction graphs, achieving competitive performance on standard benchmarks (ETH/UCY, SDD) with linear computational complexity.
Forecasting pedestrian motion has always been fundamental for autonomous navigation in crowded environments. While attention-based methods achieve strong performance, they suffer from quadratic computational complexity in modeling social interactions, limiting scalability. Additionally, the existing methods often achieve high accuracy on prediction benchmarks at the individual level, but fail to fully capture the natural movement behaviors of crowds in real-world scenarios, particularly group structures. In this study, we propose Social Graph Mamba (SGM), a novel architecture that replaces attention-based social reasoning with Selective State Space Models (SSMs) operating on dynamically constructed interaction graphs. SGM introduces a dynamic interaction graph with social triplet factorization to decompose crowd interactions sequentially, and a community-aware module to effectively discover group structures via differentiable MinCut optimization and conditions both the embedding space and multi-modal decoder on group membership. Our experiments on standard benchmarks (ETH/UCY, SDD) demonstrate competitive performance with linear sequence complexity compared to quadratic attention-based methods. We further validate SGM in physical robot experiments by integrating predicted trajectories into a Social Force Model (SFM) for real-world implementation.