Selective Social-Interaction via Individual Importance for Fast Human Trajectory Prediction

2025-06-23

AI Summary

This paper addresses the problem of redundant neighborhood interference degrading both accuracy and efficiency in crowd interaction modeling for trajectory prediction. To tackle this, we propose a differentiable dynamic neighbor selection method. Its core innovation is an importance estimation module jointly optimized with Gumbel-Softmax to enable end-to-end trainable selection of key pedestrians—balancing interpretability, selection rationality, and computational efficiency. Unlike conventional approaches relying on hand-crafted neighborhood radii or fixed neighbor counts, our method adaptively identifies contextually relevant agents. Evaluated on the JRDB dataset, it achieves a ~2.3× speedup over baseline models while maintaining state-of-the-art trajectory prediction accuracy—achieving competitive or superior ADE/FDE performance relative to leading methods. This work establishes a new paradigm for efficient and interpretable crowd interaction modeling, advancing both practical deployment and model transparency.

Technology Category

• Intelligent Robots: State Estimation
• Planning, Routing, and Scheduling: Model-Based Reasoning
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

This paper presents an architecture for selecting important neighboring people to predict the primary person's trajectory. To achieve effective neighboring people selection, we propose a people selection module called the Importance Estimator which outputs the importance of each neighboring person for predicting the primary person's future trajectory. To prevent gradients from being blocked by non-differentiable operations when sampling surrounding people based on their importance, we employ the Gumbel Softmax for training. Experiments conducted on the JRDB dataset show that our method speeds up the process with competitive prediction accuracy.

Problem

• Select important neighbors for human trajectory prediction
• Estimate individual importance for accurate trajectory forecasting
• Speed up prediction while maintaining competitive accuracy

Innovation

• Importance Estimator selects key neighbors
• Gumbel Softmax enables differentiable sampling
• Faster prediction with competitive accuracy