TACET: Context-Appropriate Acoustic-Social Navigation for Quadrupeds

2026-10-02

AI Summary

This study addresses the lack of acoustic social awareness in quadruped robots operating within sensitive environments, where simultaneously achieving spatial avoidance and noise control remains challenging. To this end, we propose TACET, a method that pioneers the integration of acoustic features into a social navigation framework. TACET leverages vision-language models to infer social contexts and generate behavioral tokens, coupling slow reasoning with fast control through a single compact token representation. Combined with structured out-of-view memory, it jointly optimizes path planning and low-noise gaits. Real-robot experiments demonstrate that TACET reduces noise by 9.3 dBA at equivalent speeds while maintaining 100% personal space compliance, keeping acoustic intrusion below 2.9 dBA. These results confirm that TACET effectively achieves dual spatial and acoustic social appropriateness for legged robots navigating human-centric settings.

Abstract

Quadruped robots entering hospitals, care homes, and quiet offices must be context-appropriate not only in where they move but in how loudly they move: a legged robot's locomotion noise, dominated by foot-ground impacts, is itself a social variable. Prior social navigation respects human space but treats the robot as acoustically uniform, while quiet-locomotion methods reduce noise to an operator-specified, context-blind level. We present TACET, a context-appropriate acoustic-social navigation method that infers social context from the robot's egocentric view and decides both where it walks and how loudly, coupling a slow fine-tuned vision-language reasoner to a fast reactive controller through a single compact behavior token,. The same token conditions both a social costmap (where to go) and a quiet locomotion policy (how loudly to move), while a structured out-of-view memory keeps recently seen people in the reasoner's context after they leave the camera view. On a real quadruped, context-conditioned locomotion lowers locomotion noise by up to 9.3 dBA at matched speed, and across our scenarios the full method keeps personal-space compliance at 100% with low acoustic intrusion (<=2.9 dBA), jointly improving spatial and acoustic performance in the evaluated scenarios. The project page is available at https://rcilab.khu.ac.kr/tacet/.

Problem

• Social Navigation
• Acoustic-aware Locomotion
• Quadruped Robots
• Context Awareness

Innovation

• Acoustic-Social Navigation
• Vision-Language Reasoner
• Behavior Token
• Quiet Locomotion Policy
• Out-of-View Memory