Salvato in:
Dettagli Bibliografici
Autori principali: Li, Jingyu, Liu, Zhe, Wu, Wenxiao, Zhang, Li
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:https://arxiv.org/abs/2605.19594
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916027600732160
author Li, Jingyu
Liu, Zhe
Wu, Wenxiao
Zhang, Li
author_facet Li, Jingyu
Liu, Zhe
Wu, Wenxiao
Zhang, Li
contents Navigating to instance-level targets in complex environments is a challenging problem. Many existing zero-shot methods achieve strong performance by modeling the entire environment and leveraging large language models for scene understanding. However, such strategies primarily focus on exploring new regions while lacking a deeper exploitation of information from previously explored areas. Consequently, when targets are missed or misidentified within previously visited regions, navigation failures occur frequently. To address these limitations, we propose MCNav, a memory-aware navigation framework with a dynamic cognitive map. This map stores efficiently queryable information about relevant objects in explored areas. Building on this memory structure, MCNav introduces two memory-aware exploration strategies: goal re-validation, which re-assesses previously seen objects to correct matching failures, and missed goal re-exploration, which estimates the likelihood that a target is present in an explored region from contextual cues. These strategies are further stabilized by a blacklist mechanism to prevent repeated errors and a double-check mechanism for high-confidence confirmation. We evaluate MCNav on the HM3Dv1 and HM3Dv2 datasets across three different tasks, where it achieves state-of-the-art performance, particularly on the instance-level goal navigation task.
format Preprint
id arxiv_https___arxiv_org_abs_2605_19594
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle MCNav: Memory-Aware Dynamic Cognitive Map for Zero-shot Goal-oriented Navigation
Li, Jingyu
Liu, Zhe
Wu, Wenxiao
Zhang, Li
Robotics
Navigating to instance-level targets in complex environments is a challenging problem. Many existing zero-shot methods achieve strong performance by modeling the entire environment and leveraging large language models for scene understanding. However, such strategies primarily focus on exploring new regions while lacking a deeper exploitation of information from previously explored areas. Consequently, when targets are missed or misidentified within previously visited regions, navigation failures occur frequently. To address these limitations, we propose MCNav, a memory-aware navigation framework with a dynamic cognitive map. This map stores efficiently queryable information about relevant objects in explored areas. Building on this memory structure, MCNav introduces two memory-aware exploration strategies: goal re-validation, which re-assesses previously seen objects to correct matching failures, and missed goal re-exploration, which estimates the likelihood that a target is present in an explored region from contextual cues. These strategies are further stabilized by a blacklist mechanism to prevent repeated errors and a double-check mechanism for high-confidence confirmation. We evaluate MCNav on the HM3Dv1 and HM3Dv2 datasets across three different tasks, where it achieves state-of-the-art performance, particularly on the instance-level goal navigation task.
title MCNav: Memory-Aware Dynamic Cognitive Map for Zero-shot Goal-oriented Navigation
topic Robotics
url https://arxiv.org/abs/2605.19594