podomy/weld
Hardware & IoTGo robotics middleware wire-compatible with ROS 2, aimed at robot fleets that need coordination over unreliable links.
What it is
Weld is a robotics middleware written in Go. It runs a leaderless data plane that carries ROS 2 topics, services, actions and parameters, and is wire-compatible with ROS 2 so it can sit in existing ROS 2 setups. Version one ships with a CycloneDDS backend. The target use case is robot fleets that must keep coordinating when network links drop.
The project's main claim is performance: up to 68% lower latency than ROS 2 on the same host, and zero packet loss to 100 kHz where ROS 2 first drops at 10 kHz. These figures come from the project's own 110-row benchmark matrix in testbed/bench, which is not reproduced here as an independent measurement.
Getting started
The README does not publish install commands, a quick-start example, or a first-run walkthrough. The only documented entry points are the docs/ directory in the repository and the Go module manifest (go.mod), so the practical starting point for a Go user is the module itself, with the docs as the reference for the data plane and ROS 2 compatibility.
Given the project states it accepts no pull requests or GitHub Issues, questions about setup or integration need to go through email rather than the repository tracker.
Constraints
- Language and runtime: the implementation is in Go; the current backend is CycloneDDS, so DDS is part of the deployment picture.
- Compatibility: it claims ROS 2 wire compatibility, but that compatibility is stated for the current version and its CycloneDDS backend.
- License: AGPL-3.0 or later. Teams embedding Weld in a network service should review the AGPL obligations before shipping.
- Contribution model: no pull requests or GitHub Issues; proposals go by email as an RFC first, then a
git format-patchseries.
HotRepo's take
Weld entered HotRepo's new-project snapshot on 2026-11-20 with 101 stars gained in roughly four hours and 1 new fork, but no GitHub daily rank has been recorded. That is a single short observation window, not evidence of sustained adoption, and the near-absent fork count means there is no visible external code activity to cross-check against. The project's headline is a latency and loss benchmark against ROS 2, yet the repository ships without public install instructions or an issue tracker, so the benchmark should be treated as the author's claim until independently reproduced.
Written by AI from the project's README, updated . The project's own docs are the source of truth.
You might also like
- ros-navigation/navigation2A ROS 2 navigation framework that gives mobile robots mapping, localization, path planning and autonomous movement.
- google-deepmind/robotics-antigravity-pluginAgent skills that let Gemini-based coding agents build MuJoCo simulations and control real robot arms.
- intrinsic-ai/intrinsic-coreA local runtime, SDK, and real-time control framework for building industrial robotics applications on ROS 2.
- mozihe/z-localizationA ROS 2 LiDAR-inertial localization stack for mobile robots, with mapping, relocalization and UWB assistance.
- rerun-io/rerunOpen-source SDK and viewer for logging, visualizing, and querying multimodal robotics and sensor data over time.