[2607.16636] PhyAgentOS: A Self-Evolving Operating System for Embodied Agents with Decoupled Cognitive Planning and Physical Execution[arxiv.org/pdf/2607.16636] Our technical report has been published. Stay tuned!

“Sort everything into three plates.” 7 objects. No categories specified. And a grasp that misses. Watch PhyAgentOS connect task reasoning, skill composition and execution feedback—from a vague instruction to a sorted table, with corrections along the way.

Different robots, one unified physical interface From robot arms to more physical configurations, from task execution to simulation evaluation — connecting a diverse physical world through a unified interface.

PhyAgent OS: Three control paradigms. One embodied system. 👇 See how three control paradigms execute the task.

⚡️ PAOS Integrates JEV: Unlocking Real-Time Capabilities for Embodied AI Harnesses ，4.3x latency reduction compared to Astra, zero performance loss 🧵🌐 phy-agent-os.x-era.com/ 🔗 github.com/PhyAgentOS/PhyAge…

[RoboCasa365 Evaluation]Demonstrates the RoboCasa365 Target50 evaluation integration: launching Verification and the π0.5 policy service, issuing refrigerator-closing tasks through natural language, and recording auditable execution results.

[CALVIN Evaluation]Demonstrates the integration workflow for CALVIN ABC→D long-horizon task evaluation: launching the π0.5 FlowSDE policy service, creating evaluation sessions through natural language, and using the verifier to track execution and recovery results.

[LIBERO Evaluation]Demonstrates the session-based evaluation workflow of the OpenVLA Agent on the LIBERO benchmark: launching TargetWS and OpenPI Policy, generating task sessions, scheduling execution, and recording results.

[Real-World Task Chain]Humanoid/mobile robots continuously perform physical tasks in real-world rooms, including shopping, organizing, pick-and-place, bag opening, and storage, demonstrating reliable execution from planning to action.

[deployment]From device connection, virtual environment, configuration files, and API keys to issuing the command for Go2 to stand up and walk, this demonstrates the entire path of first-time deployment and runtime execution.

Hackathon @Beta_ucb recap ⚡️ Built a bunch. Learned a ton. Took home robot swag from the challenge 🏆 10/10 would do it again 🤖

If you’ve ever worked on robotics or embodied AI, you know this pain: Paper: understood ✅ Environment: set up ✅ Real robot: ❌ — moves once… then dies. Then the loop begins: driver issues, version conflicts, ROS topics mismatch, reinstall everything… A whole day gone, progress ≈ zero. Switch to another robot? Start over. It’s not you. It’s the stack. We’ve been working on something to fix this: 👉 PhyAgentOS An open-source system designed to pull developers out of “debugging hell” and bring the focus back to intelligence itself. What it changes: ⚡ From days to hours to get a robot running 🔁 Switch across robots without rewriting the whole stack 🤝 Multi-robot collaboration without chaos 🔍 Fully transparent decision process (no more black boxes) In short: less glue code, more actual intelligence. Our vision is simple: Stop turning robotics into an endless loop of “tune model → fix drivers → patch environment” Instead: 👉 Build structured, understandable intelligent systems 👉 Enable robots to collaborate, learn, and evolve over time 👉 Lower the barrier so more developers can actually build in the real world It’s fully open source — contributions welcome: web page：phy-agent-os.net Github：github.com/PhyAgentOS/PhyAge… If you’ve ever felt: “the paper works, but the robot doesn’t” This might be for you. #robotics #embodiedAI #AIagents #multiAgentSystems #roboticsOS #ROS #robotLearning #physicalAI #autonomousRobots #openSourceRobotics #AIroboticsDemo #robotControl #LLMrobotics #VLAmodel #simulationToReal #roboticsFramework

Paper works. Robot doesn’t. If you’ve built robots, you know the pain: drivers break, ROS topics mismatch, everything works… until it doesn’t. We built PhyAgentOS to fix this. 👉 Hours to deployment, not days 👉 Multi-robot coordination out of the box 👉 Fully open-source Web Page：phy-agent-os.net Github：github.com/PhyAgentOS/PhyAge… #robotics #embodiedAI #AIagents #multiAgentSystems #roboticsOS #ROS #robotLearning #physicalAI #autonomousRobots #openSourceRobotics #AIroboticsDemo #robotControl #LLMrobotics #VLAmodel #simulationToReal #roboticsFramework