SAGE: A Socially-Aware Generative Engine for Heterogeneous Multi-Agent Navigation

2026-07-17

AI Summary

This work addresses the challenge of safe navigation and compliance with social norms in open human-robot environments where heterogeneous agents coexist. Existing approaches struggle with asymmetric interactions, tightly coupled prediction-planning loops, and soft social constraints. To overcome these limitations, we propose the first unified prediction-planning framework that integrates a heterogeneous graph Transformer, a diffusion generative model, and a training-free energy-guided mechanism. Our method models type-specific interactions via directed heterogeneous graphs, jointly generates environmental trajectories and robot paths using diffusion models, and explicitly incorporates collision avoidance, kinematic constraints, task objectives, and role-conditioned social norms through differentiable energy terms during inference. Experiments on ETH/UCY and SDD benchmarks demonstrate significant reductions in collision and social violation rates, scalability to fleets of up to 20 robots, and seamless adaptation to diverse normative requirements without retraining, effectively balancing safety, task performance, and social compliance.

Abstract

Safe and socially compliant navigation in open human-robot environments requires robots to reason about heterogeneous participants with different dynamics, autonomy levels, and social roles. Existing trajectory prediction and planning methods often rely on homogeneous interaction assumptions or enforce only geometric collision constraints, making it difficult to jointly model asymmetric interactions, coupled prediction-planning, and soft social norms. This paper proposes SAGE, a socially-aware generative engine for heterogeneous multi-agent navigation. SAGE represents robots and surrounding entities as a directed heterogeneous graph and employs a Heterogeneous Graph Transformer (HGT) to encode type-specific asymmetric interactions. Conditioned on the resulting context, a diffusion-based generative module jointly models future entity trajectories and robot trajectory plans. During inference, a training-free safety-social energy guidance mechanism refines sampled robot trajectories using differentiable collision, kinematic, task-progress, and role-conditioned social-compliance terms. Extensive experiments on real-world (ETH/UCY and SDD) and synthetic datasets verify the effectiveness of SAGE in improving safety and social compliance while maintaining task performance. The proposed guidance mechanism consistently reduces collision and social-violation rates, scales to teams of up to 20 robots, and enables explicit control of the safety-accuracy-task trade-off without retraining. These findings demonstrate the potential of SAGE as a scalable framework for socially-aware multi-agent navigation in complex environments.

Problem

• heterogeneous multi-agent navigation
• socially-aware navigation
• asymmetric interactions
• social compliance
• trajectory prediction and planning

Innovation

• heterogeneous multi-agent navigation
• socially-aware planning
• diffusion-based trajectory generation
• heterogeneous graph transformer
• energy-guided inference