Visual Language Maps for Robot Navigation

By Chen Huang, Oier Mees, Andy Zeng, Wolfram Burgard. 2022

618 citations recorded by the source corpus.

Grounding language to the visual observations of a navigating agent can be performed using off-the-shelf visual-language models pretrained on Internet-scale data (e.g., image captions). While this is useful for matching images to natural language descriptions of object goals, it remains disjoint from the process of mapping the environment, so that it lacks the spatial precision of classic geometric maps. To address this problem, we propose VLMaps, a spatial map representation that directly fuses pretrained visual-language features with a 3D reconstruction of the physical world. VLMaps can be autonomously built from video feed on robots using standard exploration approaches and enables natural language indexing of the map without additional labeled data. Specifically, when combined with large language models (LLMs), VLMaps can be used to (i) translate natural language commands into a sequence of open-vocabulary navigation goals (which, beyond prior work, can be spatial by construction, e.g., “in between the sofa and the TV” or “three meters to the right of the chair”) directly localized in the map, and (ii) can be shared among multiple robots with different embodiments to generate new obstacle maps on-the-fly (by using a list of obstacle categories). Extensive experiments carried out in simulated and real-world environments show that VLMaps enable navigation according to more complex language instructions than existing methods. Videos are available at https://vlmaps.github.io.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Multimodal Learning, Robotics & Control

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Multimodal Learning

In Robotics & Control

Citations in the corpus

PaLM-E: An Embodied Multimodal Language Model (2023, background): “…(Huang et al., 2022c), generating world models (Nottingham et al., 2023; Zellers et al., 2021a), planning over graphs and maps (Shah et al., 2022; Huang et al., 2022a), visual explanations (Wang et al., 2023), program generation (Liang et al., 2022; Singh et al., 2022), or injecting information…”
OpenMask3D: Open-Vocabulary 3D Instance Segmentation (2023, background): “Recent success of 2D open-vocabulary segmentation models such as LSeg [39], OpenSeg [18], and OV-Seg[41] has motivated researchers in the field of 3D scene understanding to explore the open vocabulary setting [6, 10, 17, 21, 27, 29, 32, 34, 45, 49, 56, 57].”
Robotic Control via Embodied Chain-of-Thought Reasoning (2024, neutral): “A number of works have explored similar techniques in the context of high-level task planning for robotics [48, 49, 50, 51, 52, 53, 54, 55, 56].”
LLM-Planner: Few-Shot Grounded Planning for Embodied Agents with Large Language Models (2022, methodology): “Existing works use LLMs as a static generator for HLP [12, 13, 22, 30, 33, 40], use LLMs as a ranker that ranks all admissible actions in the current environment [1, 22], or add ground-truth environmental input to the LLM prompt [33].”
Instruct-NeRF2NeRF: Editing 3D Scenes with Instructions (2023, background): “Instruct-Pix2Pix [2] demonstrates the effectiveness of instructions in 2D image tasks, as do other works in other domains such as robotic navigation [11].”
The rise and potential of large language model based agents: a survey (2025, positive): “Some studies [275, 276] highlighted the importance of spatial representation to obtain the precise localization of targets by leveraging the pre-trained VLM model to combine visual features from images with 3D reconstructions of the physical world [250].”
Open-vocabulary Queryable Scene Representations for Real World Planning (2022, background): “Concurrent work VLMaps [21] also explores this concept, by fusing pretrained visual-language model features into a geometric reconstruction of the scene.”
Foundation models in robotics: Applications, challenges, and the future (2023, neutral): “Another related work is VLMaps [171], which projects pixel embeddings from LSeg to grid cells in a top-down grid map.”
LangSplat: 3D Language Gaussian Splatting (2023, background): “The field of open-ended language queries in 3D has attracted increasing attention due to its various applications such as robotic navigation [14] and manipulation [35], 3D semantic understanding [9, 49] and editing [20], autonomous driving [16], and augmented/virtual reality [23].”
CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Models (2025, neutral): “…works utilize VLMs as intermediate components for perception and control, leveraging their strong semantic understanding and reasoning capabilities to decompose complex tasks [18, 23, 31, 50], detect objects [16, 22], or generate dense rewards [11, 37, 68] or goals [2, 9, 13, 41, 42, 49, 65, 74].”
NavGPT: Explicit Reasoning in Vision-and-Language Navigation with Large Language Models (2023, background): “(Huang et al. 2022) concentrates its efforts on the application of an LLM for the generation of code.”
ConceptFusion: Open-set Multimodal 3D Mapping (2023, methodology): “We refer to these baselines as LSeg-3D and OpenSeg-3D respectively .”
SpatialRGPT: Grounded Spatial Reasoning in Vision Language Model (2024, background): “…not only for visual understanding, but also for enabling practical applications in fields like robotics [10, 11] and augmented reality [12], where precise spatial awareness is crucial for tasks such as navigation [13], manipulation [10], and interaction with real-world environments [14].”
Octo: An Open-Source Generalist Robot Policy (2024, background): “Multiple works leverage diverse non-robot data or pretrained vision-language foundation models to boost policy generalization to new scenes and tasks [86, 103, 96, 16, 38, 11, 84, 36, 4, 37, 7, 3, 46, 15, 23].”
A Comprehensive Overview of Large Language Models (2023, neutral): “robotics [196], [197], [198], [199], offering significant potential to enhance a robot’s ability to navigate complex environments with precision and adaptability.”
LERF: Language Embedded Radiance Fields (2023, background): “VL-Maps [18] and Open-Scene [27] build a 3D volume of language features which can be queried for navigation tasks, by using pre-trained pixel-aligned language encoders [14, 21].”
eMEM: A Hybrid Spatio-Temporal Memory System For Embodied Agents (2026, unknown): “VLMaps [Huang et al., 2023] fuses CLIP embeddings with spatial reconstructions at the voxel level; eMEM uses the same idea at the episodic level, enabling temporal reasoning that voxel-grid representations do not support.”