Embodied AI GitHub Repositories (2026) · Open Source Tools

[IROS 26 ScaleInfra Workshop Best Tool Paper] Involving over 50 Advanced Manipulation Policies

A brain that learns from live experience. System 1 by default: perception, memory, plasticity and action. Optional System 2 for self-observation and slow corrections.

Official codebase for Spotter: Let the Embodied Model Lead, and the VLM Reflect for It

Open source evals for physical AI. Run any LLM/VLA on any arm/humanoid against any real/sim benchmark.

Robot-agent harness with a CLI and local Web console for LIBERO short evaluation

DORA (Dataflow-Oriented Robotic Architecture) is middleware designed to streamline and simplify the creation of AI-based robotic applications. It offers low latency, composable, and distributed dataflow capabilities. Applications are modeled as directed graphs, also referred to as pipelines.

2,000 curated embodied AI and robotics resources with bilingual metadata and transparent ranking

A minimal set of agent workflows and tools for robotics

The AI agent that physically operates a phone — the way you do.

Benchmarking Coding Agents for Interactive World Modeling from Embodied Videos - Rethinking Embodied Real2Sim from a Software Engineering Perspective

Physical AI runtime for embodied agents — governed action, verified experience, physical memory, and skill evolution.

A curated list of fruit fly connectome projects: MaleCNS, FlyWire, brain simulations, embodied models, games, and research tools.

preview of PhysicalRSI infra an engine

PhyAgentOS is a Recursive Self-Improving (RSI) physical agent operating system that enables agents to recursively self-improve through agentic workflows.

The Agentic Operating System for Robots

An end-to-end, GPU-accelerated, and modular platform for building generalized Embodied Intelligence.

The first exam for code agents in the physical world: frozen robot tasks from many benchmark suites, each answered with a program and verified in a separate sandbox. Runs on Harbor.

Open-source embodied agent framework: execution, feedback, and continual improvement

Turn world-model imagination into corrective training data for robot policies — a single-GPU, falsifiable pipeline (SmolVLA · LIBERO · CEM · paired simulator replays).

World In-Context Learning for Agentic Robot Manipulation