CVPR Findings2026
A Simple Framework for Visual Navigation
Faith Johnson, Bryan Bo Cao, Shubham Jain, Ashwin Ashok, Kristin Dana
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 mapbased methods fall short, prompting a shift toward learningbased approaches with minimal exploration. In this work, we develop a hierarchical framework 1 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, lightweight, 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 interpretability 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.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
- Object Goal Navigation using Goal-Oriented Semantic Exploration cites Learning To Explore Using Active Neural SLAM
- Object Goal Navigation using Goal-Oriented Semantic Exploration cites Habitat: A Platform for Embodied AI Research
- Learning To Explore Using Active Neural SLAM cites Habitat: A Platform for Embodied AI Research
Builds on20
- Habitat: A Platform for Embodied AI ResearchManolis Savva, Jitendra Malik, Devi Parikh, Dhruv Batra et al.ICCV 2019 · 1,863 citations
- A Minimalist Approach to Offline Reinforcement LearningScott Fujimoto, Shixiang Shane GuNeurIPS 2021 · 1,292 citations
- Object Goal Navigation using Goal-Oriented Semantic ExplorationDevendra Singh Chaplot, Dhiraj Gandhi, Abhinav Gupta, Ruslan SalakhutdinovNeurIPS 2020 · 857 citations
- Learning To Explore Using Active Neural SLAMDevendra Singh Chaplot, Dhiraj Gandhi, Saurabh Gupta, Abhinav Gupta et al.ICLR 2020 · 603 citations
- Goal-Conditioned Reinforcement Learning with Imagined SubgoalsElliot Chane-Sane, Cordelia Schmid, Ivan LaptevICML 2021 · 183 citations
Related papers
- Visual Graph Memory with Unsupervised Representation for Visual NavigationObin Kwon, Nuri Kim, Yunho Choi, Hwiyeon Yoo et al.ICCV 2021 · 86 citations
- Learning Navigational Visual Representations with Semantic Map SupervisionYicong Hong, Yang Zhou, Ruiyi Zhang, Franck Dernoncourt et al.ICCV 2023 · 56 citations
- Implicit Obstacle Map-driven Indoor Navigation Model for Robust Obstacle AvoidanceWei Xie, Haobo Jiang, Shuo Gu, Jin XieACM MM 2023 · 9 citations
- The Surprising Effectiveness of Visual Odometry Techniques for Embodied PointGoal NavigationXiaoming Zhao, Harsh Agrawal, Dhruv Batra, Alexander G. SchwingICCV 2021 · 50 citations
- AwareVLN: Reasoning with Self-awareness for Vision-Language NavigationWenxuan Guo, Xiuwei Xu, Yichen Liu, Xiangyu Li et al.CVPR 2026 · 7 citations
