# InternRobotics/InternNav

InternRobotics' open platform for building generalized navigation foundation models.

Repository: https://github.com/InternRobotics/InternNav
Canonical: https://ross.abutalabs.com/products/internnav
Homepage: https://internrobotics.github.io/user_guide/internnav/index.html
Language: Jupyter Notebook
License: MIT
License Family: permissive
Topics: mllms, navigation, robotics, vision-language-action-model, vision-language-navigation, visual-navigation, spatial-ai, spatial-intelligence, vla, vlm
Last push: 2026-03-10T01:37:19+00:00

## Health v2 (maintenance only)
Score: 58/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 71, release rhythm 58, longevity 28
- inputs: {"age_days": 405, "days_push": 177, "days_rel": 203, "gap_med": 31.5, "n_releases_24m": 5}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1061, forks 144 (observed 2026-08-28T04:03:25.919045+00:00)

## What it is
InternNav is an open-source PyTorch-based toolbox for building embodied navigation foundation models, supporting vision-language navigation (VLN-CE), visual navigation, and dual-system VLA models like InternVLA-N1. It integrates with Habitat and Isaac Sim simulators and ships datasets, benchmarks, and 10+ baselines for training and evaluating navigation agents.

## Use cases
- train a vision-language navigation model
- evaluate VLN agents in Habitat or Isaac Sim
- build a navigation foundation model for robots
- fine-tune InternVLA-N1 on custom navigation data
- benchmark embodied navigation baselines
- collect demonstration datasets for robot navigation
- zero-shot visual navigation in the real world

## When to choose
- you need a modular end-to-end navigation research stack with simulators, datasets, and baselines included
- you want to train or evaluate state-of-the-art VLN or dual-system navigation models
- you need compatibility with both Habitat and Isaac Sim

## When to avoid
- you need simple point-to-point path planning without learning-based models
- you lack GPU resources for training large VLA models
- you need a production robot navigation stack rather than a research toolbox

## Facets
- artifact type: framework
- maturity: active
- function: machine-learning, deep-learning, simulation, robotics, computer-vision, nlp, llm-training, benchmarking, data-science
- domain: robotics, machine-learning, deep-learning, artificial-intelligence, simulation, autonomous-vehicles
- platform: python
- tags: vision-language-navigation, vision-language-action-model, embodied-ai, vln, habitat, isaac-sim, navigation-foundation-model, pytorch, spatial-intelligence, mllm, linux, gpu, docker

## Member repositories
- InternRobotics/InternNav (main) score 58

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:25.919045+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:56:48.127996+00:00, confidence not recorded.
  - readme: https://github.com/InternRobotics/InternNav (fetched 2026-08-28T04:03:25.919045+00:00, sha 19967e44ce4c)
  - homepage: https://internrobotics.github.io/user_guide/internnav/index.html (fetched 2026-08-29T12:58:43.799470+00:00, sha 3857df0bd97a)
- Data as of 2026-08-30T08:39:29.467469+00:00.