Microduck is the control software for a tiny biped robot that uses reinforcement learning policies to walk, roll, and interact with its environment.

A collection of Reinforcement Learning robotic environments powered by the MuJoCo physics engine and the Gymnasium API for training AI agents in manipulation and navigation tasks.

RL training environments for the Microduck bipedal robot that use high-fidelity actuator modeling and domain randomization to enable seamless sim2real transfer of complex behaviors.

An AI-powered 3D design layer that generates editable parametric CAD models for hardware enclosures and assemblies from text and image requirements.

A library for creating pixel-based gridworld environments where large numbers of agents interact in competitive scenarios to study collective intelligence.

A set of communication-oriented multi-agent environments for training AI agents to coordinate, communicate, and compete in particle-based simulations.

An open-source, DIY robot vacuum cleaner built with Raspberry Pi and ROS2 that uses 2D LiDAR for autonomous local navigation without cloud dependence.

A library of agent skills that enables AI to generate, inspect, and prepare CAD models and robot-description files for simulation and fabrication.

Mobile MCP is an open‑source MCP server that lets AI agents control iOS/Android simulators, emulators, and real devices through a single API. It uses the native accessibility tree for fast, structured UI interaction, offers full device control (taps, swipes, app install, screen recording, etc.), and integrates with Claude, Gemini, Copilot, Codex and any MCP‑compatible client. Install via `npx @mobilenext/mobile-mcp@latest`, configure with a tiny JSON snippet, and run locally or as an SSE HTTP server.

A reverse-engineering project that reconstructs the assembly drawings, CAD models, and hardware specifications of the Microduck bipedal robot from its official simulation files.

Zetta is an efficient closed-loop harness for embodied AI that evolves code-based critics and recovery skills online to improve robot task success rates without retraining the base policy.

Cua is an open‑source stack that gives AI agents the ability to control desktop applications in the background across macOS, Windows, Linux and Android. It includes cross‑OS drivers, a Python sandbox SDK for spinning up VMs/containers, a benchmark suite (Cua‑Bench) for evaluating computer‑use agents, and Lume for fast macOS virtualization on Apple Silicon. The project is MIT‑licensed and targets researchers and developers building autonomous agents that need to interact with real GUIs.

A PyTorch-native library providing models, datasets, and tools to standardize real-world robotics, lowering the barrier to entry for training and deploying AI policies.

An open-source ESP32-based AI voice chatbot framework that integrates LLMs and the Model Context Protocol (MCP) for real-time voice interaction and hardware control.

An agent framework for embodied AI that uses a decoupled execution gateway and evidence-based verification to ensure robots achieve their goals and recursively improve from experience.

PageIndex is an open‑source RAG system that replaces vector‑store retrieval with a hierarchical tree index built from a document’s natural sections. An LLM reasons over this tree to fetch answers, providing explicit page‑level citations. It works locally via a Python SDK (`pip install pageindex`) or in the cloud for OCR and massive corpora, supports any LiteLLM‑compatible model, and claims state‑of‑the‑art accuracy on FinanceBench.

A high-performance physics engine for the fast and accurate simulation of articulated structures, widely used in robotics and machine learning research.

A unified simulation platform for dexterous manipulation research featuring a contact-first physics engine, robotics SDK, and RL-ready simulation harness.

A lightweight, GPU-accelerated framework for robot learning that combines Isaac Lab's manager-based API with MuJoCo Warp.

A comprehensive learning library for Embodied AI that guides users from basic simulation to the reproduction of state-of-the-art Vision-Language-Action (VLA) models and real-world robot deployment.

The repository provides an open‑source, 3‑D‑printable robot arm (SO‑100 / SO‑101) with a full bill‑of‑materials, printable STL files, assembly guides, and ready‑made integration with the 🤗 LeRobot library for AI‑driven tele‑operation and reinforcement‑learning experiments.

An open-source quadruped robot project based on the ESP32, featuring 3D-printable parts, an emotive OLED face, and a suite of animation and simulation tools.

BAM provides high-fidelity servo actuator models and an identification pipeline to improve the accuracy of robotic simulations, specifically for better Sim-to-Real transfer in Reinforcement Learning.

An AI-assisted PCB design plugin for KiCAD 10 that allows AI assistants to design schematics and layout PCBs using the Model Context Protocol.

A codebase for humanoid whole-body control featuring SONIC, a foundation model that enables robots to perform a wide range of natural motor skills learned from large-scale human motion data.

A kinematic motion diffusion model that generates controllable 3D human and robot motions using text prompts and kinematic constraints.

An end-to-end pipeline that converts human demonstration videos into 3D simulation assets and physics-grounded robot training data for policy learning.

Kairos is a 4B-parameter cross-embodiment world model that unifies video generation and executable action prediction for real-time robotic control across diverse hardware platforms.

A genuine open‑source toolkit for recording, merging, auditing, cleaning and uploading multimodal robot manipulation datasets (RGB and RGB‑D) in LeRobot format, with optional semantic evaluation via a pretrained VLA policy.

ClawBench is an open‑source benchmark that evaluates AI browser agents on 281 real‑world web tasks (e.g., booking travel, ordering food). It provides task specs, a multi‑layer trace recorder, and an LLM judge (deepseek‑v4‑pro) to score success. Install via `pip install clawbench-eval`, configure your model and judge API key, then run tasks in a Docker/Podman‑based Chromium harness. Results are posted to a public leaderboard and the full data are available on Hugging Face.

A simulation platform for physical AI and robotics that combines a multi-physics engine, photo-realistic rendering, and a cross-platform compiler for scalable research.

An open-source, expressive robot platform for AI builders featuring Python/JS SDKs and an app store integrated with Hugging Face Spaces.

An open-source research platform for developing Vision-Language-Action (VLA) models with a modular, plug-and-play codebase for generalist robotics.

NVIDIA Isaac GR00T N1.7 is a vision‑language‑action model that turns video + text instructions into continuous robot actions. It uses a Qwen3‑VL visual backbone, a diffusion transformer head, and a relative end‑effector action space to work across many robot embodiments. The repo provides installation scripts, data‑format specs, zero‑shot inference, fine‑tuning tools, benchmark evaluations, and deployment pipelines (ONNX/TensorRT) for platforms ranging from desktop GPUs to Jetson devices.

An open-source operating system for robotics that upgrades driver assistance systems in over 300 supported cars.

A GPU-accelerated open-source framework built on NVIDIA Isaac Sim for simplifying robotics research in reinforcement learning, imitation learning, and motion planning.

A Python code collection and textbook of robotics algorithms covering localization, mapping, SLAM, path planning, and path tracking.

ArduPilot is a comprehensive open-source autopilot software suite that provides advanced control for a variety of unmanned vehicles, including drones, planes, and submarines.

OpenCatESP32 is the firmware that runs Petoi’s open‑source quadruped robots (Bittle X, Nybble Q, and the upcoming Quaddle) on an ESP32‑based BiBoard. It provides real‑time gait control, a motion‑recording “Puppet Mode”, and enough processing power to add vision or reinforcement‑learning experiments, all programmable via the Arduino IDE.

The Simulink Agentic Toolkit lets LLM coding agents (Claude, Copilot, Codex, Gemini, etc.) interact with Simulink models through a MATLAB MCP Server. It provides ready‑made tools (e.g., model_overview, model_edit) and a catalog of skills that encode Model‑Based Design best practices. An automated installer sets up the MCP server, registers the tools, and adds the skills to the chosen agent. After a one‑time `satk_initialize` in MATLAB, the agent can read, edit, test, and validate Simulink models safely, with optional support for custom block libraries.

LingBot-VLA 2.0 is a Vision-Language-Action foundation model that enables robots to generalize across different embodiments and tasks using a unified action representation and MoE action experts.

FastWAM is a high-performance World Action Model for robotics that enables fast action prediction by optionally skipping test-time future imagination, significantly reducing inference latency.

An AI-native automation layer for EasyEDA Pro that enables AI agents to automate schematic capture and PCB layout through a type-safe action system and a verified circuit block library.

Newton is a GPU-accelerated, differentiable physics simulation engine built on NVIDIA Warp and MuJoCo Warp, designed for scalable robotics research.

An open-source simulator for drones and cars built on Unreal Engine, designed as a platform for AI research in deep learning, computer vision, and reinforcement learning.

An AR diffusion framework for robot control that unifies visual world modeling and action inference to improve long-horizon performance and generalization.

An agent-native operating system and SDK for generalist robotics that allows developers to build physical applications for humanoids, quadrupeds, and drones using Python.

An open-source autopilot stack for drones and unmanned vehicles that provides a modular flight control system supporting various airframes and hardware.