Repolode
Submit
rerun-io

rerun-io/rerun

Open-source Rust toolkit for logging, visualizing, querying and streaming multimodal robotics and sensor data over time.

Open-source alternative to:RViz

View on GitHubWebsite
Stars
11,610
Forks
876
Open issues
1,239
Last commit
14 hours ago

What is it?

Rerun is a data layer for robotics and physical AI. It ingests multi-rate data such as images, point clouds, transforms, time series, joint states and video from robot logs, simulators and web sources, including MCAP and .rrd files. The built-in viewer plays these streams back in sync on a shared timeline, so developers can scrub through episodes and compare sensors side by side.

The same stored data can be queried with dataframes or SQL and streamed directly into model training, without export jobs or copies. Storage is built on column-chunk formats designed for data recorded at different rates. SDKs are available for Python, Rust and C++, and the Python package bundles the viewer.

It is aimed at engineers who need a temporal debugger that shows what a robot perceived at each moment, as a replacement for pure live-visualization tools like RViz.

Key features

  • 01Visualizes images, point clouds, transforms, time series and video in sync on a shared timeline.
  • 02Imports MCAP, .rrd and LeRobot data.
  • 03Supports querying recorded data with dataframes or SQL.
  • 04Provides SDKs for Python, Rust and C++.
  • 05Lets you go back in time with an adjustable in-memory buffer.
  • 06Runs the viewer in a browser or as a local binary.

Use cases

  • Debugging a robot by reviewing camera, lidar and map data frame by frame.
  • Recording test runs and comparing sensor streams across episodes.
  • Building training datasets by querying recorded multimodal data.
  • Replacing RViz for long-running or memory-constrained visualization.

Quick start

  1. 1

    Install the Python SDK (includes the viewer)

    pip install rerun-sdk
  2. 2

    Log data and spawn the viewer from a Python script

  3. 3

    Check the viewer binary is available

    rerun --help

Things to know

  • The project is in active development and expects breaking API changes.
  • The viewer slows down when there are too many entities, and multi-million point clouds can be slow.
  • Only the Python SDK bundles the viewer; C++ and Rust need a separate rerun install.

Topics

#robotics#visualization#multimodal-data#computer-vision#time-series#point-cloud#python#rust#cpp#multimodal

FAQ

What is rerun?
Open-source Rust toolkit for logging, visualizing, querying and streaming multimodal robotics and sensor data over time. Rerun is a data layer for robotics and physical AI. It ingests multi-rate data such as images, point clouds, transforms, time series, joint states and video from robot logs, simulators and web sources, including MCAP and .rrd files. The built-in viewer plays these streams back in sync on a shared timeline, so developers can scrub through episodes and compare sensors side by side.
Is rerun free to use? What license does it have?
rerun is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is rerun still maintained?
Actively maintained. The last commit to rerun was on Oct 19, 2026.
What can I use rerun for?
Debugging a robot by reviewing camera, lidar and map data frame by frame. Recording test runs and comparing sensor streams across episodes. Building training datasets by querying recorded multimodal data. Replacing RViz for long-running or memory-constrained visualization.
How do I install rerun?
Install the Python SDK (includes the viewer): pip install rerun-sdk Log data and spawn the viewer from a Python script Check the viewer binary is available: rerun --help
Where can I get rerun?
The source code and installation instructions are on GitHub: https://github.com/rerun-io/rerun

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects