FeudalNav: A Simple Framework for Visual Navigation

2026-01-15

arXiv.org

AI Summary

In mapless, unknown, or GPS-denied environments, conventional metric map–based navigation approaches are often inapplicable. This work proposes a hierarchical visual navigation framework wherein a transferable waypoint selection network defines subgoals at the high level, while a low-level latent memory module—relying solely on visual similarity—replaces traditional graph-structured topological representations. The resulting system is lightweight, end-to-end trainable, and operates without odometry. It offers strong interpretability and supports low-intervention interactive navigation. Evaluated in Habitat AI, the method achieves performance comparable to state-of-the-art approaches, with minimal human feedback yielding substantial improvements in success rates.

Technology Category

• Computer Vision: Visual Reasoning & Symbolic Representations
• Intelligent Robots: Localization, Mapping, and Navigation
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Visual navigation for robotics is inspired by the human ability to navigate environments using visual cues and memory, eliminating the need for detailed maps. In unseen, unmapped, or GPS-denied settings, traditional metric map-based methods fall short, prompting a shift toward learning-based approaches with minimal exploration. In this work, we develop a hierarchical framework that decomposes the navigation decision-making process into multiple levels. Our method learns to select subgoals through a simple, transferable waypoint selection network. A key component of the approach is a latent-space memory module organized solely by visual similarity, as a proxy for distance. This alternative to graph-based topological representations proves sufficient for navigation tasks, providing a compact, light-weight, simple-to-train navigator that can find its way to the goal in novel locations. We show competitive results with a suite of SOTA methods in Habitat AI environments without using any odometry in training or inference. An additional contribution leverages the interpretablility of the framework for interactive navigation. We consider the question: how much direction intervention/interaction is needed to achieve success in all trials? We demonstrate that even minimal human involvement can significantly enhance overall navigation performance.

Problem

• visual navigation
• mapless navigation
• GPS-denied environments
• learning-based navigation
• unseen environments

Innovation

• hierarchical navigation
• visual similarity memory
• waypoint selection network
• odometry-free navigation
• interactive navigation