ConsistNav: Closing the Action Consistency Gap in Zero-Shot Object Navigation with Semantic Executive Control

2026-05-10

AI Summary

This work addresses the challenge in zero-shot object navigation where agents often exhibit inconsistent actions—such as lingering or abandoning the task—after detecting the target. To resolve this, the authors propose ConsistNav, a training-free framework that coordinates a finite state machine, persistent candidate memory, and stability-aware action control through a semantic execution controller, enabling consistent utilization of cross-frame semantic evidence and suppressing erratic decisions. This approach introduces, for the first time, a training-free semantic execution mechanism that integrates open-vocabulary detectors, vision-language models, and language-guided exploration. Evaluated on HM3D and MP3D benchmarks, ConsistNav achieves state-of-the-art zero-shot ObjectNav performance, improving success rate by 11.4% and success weighted by path length by 7.9% on MP3D.

Abstract

Zero-shot object navigation has advanced rapidly with open-vocabulary detectors, image--text models, and language-guided exploration. However, even after current methods detect a plausible target hypothesis, the agent may still oscillate between exploration and pursuit, or abandon the object near success. We identify this failure mode as an action consistency gap: semantic evidence is repeatedly reinterpreted at each step without persistent commitment across the episode. We introduce ConsistNav, a training-free zero-shot ObjectNav framework built around a semantic executive composed of three coordinated modules: Finite-State Executive Controller stages target pursuit through guarded semantic phases; Persistent Candidate Memory accumulates cross-frame target evidence into stable object hypotheses; and Stability-Aware Action Control suppresses rotational stagnation, ineffective pursuit, and unverified stopping. This design changes neither the detector nor the low-level planner; instead, it controls when semantic evidence should influence navigation and when it should be suppressed or revisited. We conduct extensive experiments on HM3D and MP3D, where ConsistNav achieves state-of-the-art results among compared zero-shot ObjectNav methods and improves SR by 11.4% and SPL by 7.9% over the controlled baseline on MP3D. Ablation studies and real-world deployment experiments further demonstrate the effectiveness and robustness of the proposed executive mechanism.

Problem

• zero-shot object navigation
• action consistency gap
• semantic evidence
• object pursuit
• navigation stability

Innovation

• action consistency gap
• semantic executive control
• zero-shot object navigation
• persistent candidate memory
• stability-aware action control