wpilibsuite/allwpilib

Official Repository of WPILibJ and WPILibC

What it solves

WPILib simplifies robot programming for the FIRST Robotics Competition (FRC) and FIRST Tech Challenge (FTC). It removes the need for teams to manage complex hardware-level details, allowing them to focus on writing game-specific software. The project aims to make programming accessible for beginners while remaining powerful enough for advanced teams.

How it works

The project provides a collection of core libraries—including a Hardware Abstraction Layer (HAL), camera servers, and command frameworks—that interface directly with robot control systems. It maintains parity across Java, C++, and Python, ensuring teams can use their preferred language without disadvantage. The build system uses Gradle to manage dependencies and cross-compilation for robot hardware.

Who it’s for

It is designed for students and mentors participating in FIRST Robotics competitions who need a standardized, reliable framework to control robot hardware.

Highlights

  • Provides a Hardware Abstraction Layer (HAL) to simplify interactions with robot controllers.
  • Supports multiple programming languages (Java, C++, and Python) with feature parity.
  • Includes simulation tools to test robot code on a desktop computer.
  • Offers a modular structure with specific libraries for math, networking, and command-based programming.
  • Open-source and licensed under the BSD 3-clause license.

Related

  • Project
    TommyZihao/vlm_arm

    A framework for controlling robotic arms using multimodal large language models, allowing robots to interpret visual and verbal instructions to execute physical tasks.

  • Project
    X-Square-Robot/wall-x

    An open-source framework for embodied foundation models that enables robots to perform complex physical manipulations by bridging vision-language understanding with action prediction.

  • Project
    air-embodied-brain/Zetta-Embodiment

    An efficient closed-loop harness for embodied AI that evolves code-based recovery skills and runtime critics to improve robot task success rates without retraining base policies.

  • Project
    thu-ml/Motus

    Motus is a research‑grade codebase that combines a 5 B video‑generation model, a 2 B vision‑language model, and two smaller experts into a Mixture‑of‑Transformers world model. It learns latent robot actions from optical‑flow “delta‑pixels” and can be run in several modes (world‑model prediction, VLA, inverse dynamics, video generation). The repo supplies three‑stage training scripts, data‑conversion tools for RoboTwin 2.0 and real‑robot logs, and Hugging Face checkpoints. On the RoboTwin 2.0 benchmark it reports 87 % success, outperforming prior baselines.