Visual Language Maps for Robot Navigation

Abstract

Grounding language to the visual observations of a navigating agent can be performed using off-the-shelf visuallanguage 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.

Semantic regularization of electromagnetic inverse problems

“…Correspondingly, the encoder-decoder network needs to be modified such that it is suitable for dealing with these modified inputs and outputs (see details in Supplementary Note 1). We represent the subjects with 19-point 3D skeletons and model the indoor environment with a 3D visual-semantic map such that a given coordinate (e.g., (1.3 m, 1.2 m, 0.8 m)) can be associated with a semantic coordinate (e.g., "at the left side of the chair") 58 (see Fig. 6a and Supplementary Note 10).…”

NavGPT: Explicit Reasoning in Vision-and-Language Navigation with Large Language Models

“…A handful of contemporary studies, however, have begun to explore the utilization of generative models for navigation. Shah et al (Shah et al 2023) employs GPT-3 (Brown et al 2020) in an attempt to identify "landmarks" or subgoals, while Huang et al (Huang et al 2022) concentrates its efforts on the application of an LLM for the generation of code. Zhou et al (Zhou et al 2023) use LLM to extract the commonsense knowledge of the relations between targets and objects in observations to perform zero-shot object navigation (ZSON) (Gadre et al 2022;Majumdar et al 2022).…”

Neural Implicit Vision-Language Feature Fields

“…Large language models have been explored as an approach to high-level planning [14]- [18] and scene understanding [19], [20]. Vision-language models embedding image features into the same space as text have been applied to open vocabulary object detection [16], [17], natural language maps [15], [17], [21]- [23], and for language-informed navigation [24]- [26].…”

“…Such methods decompose the scene into a graph where edges model relations between parts of the scene. The geometry of the parts are typically represented as a signed distance functions stored in a voxel grid [15].…”