From Obstacles to Etiquette: Robot Social Navigation with VLM-Informed Path Selection

2026-02-09

IEEE Robotics and Automation Letters

AI Summary

This work proposes a socially aware navigation framework that integrates geometric path planning with a lightweight vision-language model (VLM) to address the limitations of conventional robotic navigation methods, which often prioritize obstacle avoidance while neglecting social norms and thereby disrupting human activities. The approach first generates multiple geometrically feasible paths and then employs a VLM—fine-tuned via distillation from a large foundational model on social commonsense—to evaluate and select, in real time, the path most aligned with contextual social conventions. Evaluated across four representative social scenarios, the method significantly outperforms baseline approaches, achieving the shortest duration of personal space violations, minimal time spent facing pedestrians head-on, and complete avoidance of intrusions into designated social zones, thereby enabling efficient and socially compliant navigation.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Natural Language Processing: Safety and Robustness

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Search and Retrieval-Augmented AI: Personalized, context-aware and across-device search
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Navigating socially in human environments requires more than satisfying geometric constraints, as collision-free paths may still interfere with ongoing activities or conflict with social norms. Addressing this challenge calls for analyzing interactions between agents and incorporating common-sense reasoning into planning. This paper presents a social robot navigation framework that integrates geometric planning with contextual social reasoning. The system first extracts obstacles and human dynamics to generate geometrically feasible candidate paths, then leverages a fine-tuned vision-language model (VLM) to evaluate these paths, informed by contextually grounded social expectations, selecting a socially optimized path for the controller. This task-specific VLM distills social reasoning from large foundation models into a smaller and efficient model, allowing the framework to perform real-time adaptation in diverse human-robot interaction contexts. Experiments in four social navigation contexts demonstrate that our method achieves the best overall performance with the lowest personal space violation duration, the minimal pedestrian-facing time, and no social zone intrusions. Project page: https://path-etiquette.github.io

Problem

• social navigation
• robot navigation
• social norms
• personal space
• human-robot interaction

Innovation

• social robot navigation
• vision-language model
• contextual social reasoning
• path selection
• foundation model distillation