Skip to content
RobothecaBeta
Log in

SparseVideoNav

  • AI model
  • Navigation and mapping

SparseVideoNav is a research project using sparse video generation to predict future views for beyond-the-view vision-language navigation on real robots.

  • Non-commercial
  • Non-commercial weights

Made by OpenDriveLab, Hong Kong SAR China (checked on github.com/OpenDriveLab (opens in a new tab)).

Source on GitHub: github.com/OpenDriveLab/SparseVideoNav (opens in a new tab). Facts read at commit b1a0a78 (opens in a new tab).

Licenses

Code
CC-BY-NC-SA-4.0 (opens in a new tab)

Read by Robotheca from LICENSE (opens in a new tab) (CC-BY-NC-SA-4.0) at commit b1a0a78 (opens in a new tab), on 16 Oct 2026.

Commercial use
No

The code and weights licenses do not allow commercial use.

Labels

  • Non-commercialThe code license (CC-BY-NC-SA-4.0) does not allow commercial use.
  • Non-commercial weightsThe weights license (CC-BY-NC-SA-4.0) does not allow commercial use.

What it does

  • Introduces video generation models to real-world beyond-the-view vision-language navigation.
  • Uses sparse future video generation over a 20-second horizon for longer prediction.
  • Reports sub-second trajectory inference, a 27x speed-up over an unoptimized counterpart.
  • Takes a video and a language instruction, then writes a predicted future video to an output folder.
  • Offers a command-line tool, a Gradio web demo, and a Python API for inference.
  • Requires a downloadable pipeline checkpoint placed in a models directory; a continuous action head is still pending release.

Written by Robotheca from the project's README (opens in a new tab) at commit b1a0a78.

Getting started

Difficulty
Advanced
Hardware it needs, as the project states it
Linux (tested on Ubuntu), Python 3.10, NVIDIA GPU with at least 16 GB VRAM.

Maintenance

Status
Maintained: a commit in the last 12 months
Newest commit
10 Jul 2026 on main
Latest release
None published on GitHub

Checked on GitHub on 16 Oct 2026.

Links

Robots that can run it

No robot on Robotheca is linked to it yet. Links come from a robot's robot.yaml, from its dependency files read as text, or from Robotheca's editors.

Robotheca lists AI models as a reference: nothing is hosted or run here, and every fact links to where it was read.