Robots Ask the Way: Communication-Enabled Social Navigation

2026-07-01

AI Summary

This work addresses the challenge of efficiently locating a specific target individual in multi-person environments, where existing social navigation approaches lack the ability to actively acquire information. To overcome this limitation, the authors propose CommNav, a communication-enabled social navigation framework that introduces active human–robot communication into the task for the first time, enabling the robot to query residents about the target’s location and movements using natural language. Built upon an extended Habitat 3.0c simulation platform, the navigation policy is trained with pre-trained communication tasks, LLM-generated instructions, and human-collected colloquial data. Experimental results demonstrate a 10-percentage-point improvement in Episode Success, and performance using natural language instructions shows no significant difference compared to that achieved with structured perfect information, confirming the method’s effectiveness and robustness.

Abstract

Assistive autonomous robots operating in multi-agent environments require efficient strategies to locate specific individuals among multiple residents. Current social navigation methods focus on reactive collision avoidance and trajectory adaptation, but lack mechanisms to proactively gather information through human-robot communication. We introduce Communication-enabled Social Navigation (CommNav). In this novel task, robotic agents actively seek assistance from residents to locate target individuals by requesting information about recent sightings, locations, and movements. To evaluate CommNav, we extend Habitat 3.0 to create Habitat 3.0c, a communication-enabled variant supporting multi-human environments with information exchange protocols. Adding our communication module (COMM) to a state-of-the-art social navigation model yields a 10 percentage-point improvement in Episode Success. We further investigate the transition from structured data to natural language by evaluating models trained on LLM-generated instructions and on colloquial instructions collected from a human study. Our experiments reveal that: (i) explicit human-robot communication substantially enhances multi-person navigation performance; (ii) pre-training COMM on a communication pretext task effectively addresses the challenge of occasional interaction signals; and (iii) the navigation policy is highly robust to natural, colloquial human language, achieving an episode success statistically similar to the model using perfect structured data.

Problem

• social navigation
• human-robot communication
• multi-agent environments
• target localization
• assistive robotics

Innovation

• Communication-enabled Social Navigation
• Human-Robot Communication
• Habitat 3.0c
• Natural Language Robustness
• Social Navigation