Physical AI logopeterindia.net
Knowledge Hub

Physical AI: Embodied Intelligence, Robot Foundation Models, Simulation & Robots

Physical AI is AI that perceives, reasons about and acts in the physical world — through robots, vehicles and machines. This hub links to the models, simulators, software, hardware and robots behind it, and to related directories on this site.

58 topics•8 categories•Updated October 2026

What Is Physical AI?

5 entries

Explainers and the research that frames the field.

01What Is Physical AI? (NVIDIA)NVIDIA's glossary explanation of generative physical AI and the models and data behind it. 02What Is Robot Learning? (NVIDIA)Introduction to how robots learn skills through imitation and reinforcement learning. 03Open X-EmbodimentCross-institution dataset and RT-X models for robot learning across many robot types. 04World Action ModelsHow joint video-action models let robots imagine a future before they act. 05Large Action ModelsModels and tools that translate intent into actions.

Robot Foundation Models & VLAs

7 entries

Vision-language-action models that turn perception and instructions into robot actions.

06Gemini RoboticsGoogle DeepMind's family of models that bring Gemini into the physical world for robots. 07pi0: A Vision-Language-Action Flow ModelPhysical Intelligence's generalist robot policy paper. 08openpiOpen-source models and code from Physical Intelligence, including pi0. 09NVIDIA Isaac GR00TNVIDIA's open foundation models and tools for generalist humanoid robots. 10GR00T N1 paperAn open foundation model for generalist humanoid robots. 11RT-2Vision-language-action models that transfer web knowledge to robot control. 12Multimodal Language ModelsModels that combine vision and language.

World Models & Physical-AI Platforms

5 entries

Models that learn how the world behaves and platforms for building on them.

13NVIDIA CosmosWorld foundation model platform for physical AI development and synthetic data. 14Cosmos World Foundation Model Platform paperTechnical paper describing the Cosmos platform. 15NVIDIA Cosmos on GitHubOpen models, tokenizers and tools from the Cosmos project. 16V-JEPA 2Meta's video world model for understanding, prediction and planning. 17World LabsResearch company building spatial intelligence and 3D world models.

Simulation & Synthetic Data

11 entries

Train and test robots virtually before deploying them.

18NVIDIA Isaac SimRobotics simulation and synthetic data generation built on Omniverse. 19Isaac LabUnified framework for robot learning built on Isaac Sim. 20NVIDIA OmniverseLibraries and APIs for physically based 3D simulation. 21MuJoCoMulti-Joint dynamics with Contact: a physics engine for robotics research. 22MuJoCo MenagerieCollection of high-quality robot models for MuJoCo. 23GenesisGenerative physics simulation platform for robotics and embodied AI. 24GazeboOpen-source robot simulator integrated with ROS. 25PyBullet / Bullet PhysicsPython bindings for the Bullet real-time physics engine. 26robosuiteModular simulation framework and benchmark for robot learning. 27BEHAVIORStanford benchmark for everyday household activities in simulation. 28GymnasiumStandard API for reinforcement learning environments.

Robot Software, Learning & Data

6 entries

Frameworks and datasets used to build and train robot policies.

29ROS 2Robot Operating System, the standard middleware for robot software. 30LeRobotHugging Face library with models, datasets and tools for real-world robotics in PyTorch. 31DROIDLarge-scale in-the-wild robot manipulation dataset. 32Reinforcement LearningReinforcement learning hub. 33Pose Estimation ModelsHuman keypoint detection models. 34Computer VisionComputer vision hub.

Edge Compute, Sensors & Chips

5 entries

Hardware that runs physical AI on the machine.

35NVIDIA JetsonEmbedded AI computing platform for robots and edge devices. 36AI Chip ProvidersAI accelerators and chip vendors. 37Edge AIAI at the edge. 38SensorsSensor technologies hub. 39AIoTArtificial intelligence on the Internet of Things.

Robots, Vehicles & Drones

13 entries

Companies and directories of machines that embody AI.

40FigureHumanoid robot developer. 41Agility RoboticsMaker of the Digit humanoid for industrial work. 42Boston DynamicsMaker of Spot, Stretch and Atlas robots. 431XDeveloper of humanoid home robots. 44ApptronikHumanoid robotics company. 45Unitree RoboticsMaker of quadruped and humanoid robots. 46SkydioAutonomous drones for inspection and public safety. 47WaymoAutonomous driving company. 48Advanced AI-Driven RobotsDirectory of advanced AI-driven robots. 49AI-driven Robots and DronesAI-driven robots and drones. 50Robotics PlatformsRobotics platforms compared. 51Robotics ProductsRobotics products. 52Drone TechnologiesDrone technology directory.

Related Pages

6 entries

Neighbouring directories.

53RoboticsRobotics technologies hub. 54Digital TwinsDigital twin resources. 55Digital Twins SoftwareDigital twin platforms. 56Agentic AIAgentic AI hub. 57Industrial IoT PlatformsIIoT platforms. 58Artificial IntelligenceAI hub.

Frequently Asked Questions

What is physical AI?

Physical AI is artificial intelligence that perceives, reasons about and acts in the real world, usually through robots, autonomous vehicles and smart machines, rather than only producing text or images.

What is a vision-language-action model?

A VLA model takes camera images and a language instruction and outputs robot actions, so one model can control a robot across many tasks. RT-2, pi0 and GR00T are examples.

Why is simulation important?

Robots learn faster and more safely in simulators such as Isaac Sim, MuJoCo and Genesis, with synthetic data generated at scale, before the policy is transferred to a real robot, a step called sim-to-real.

What are world models?

Models that learn how the physical world changes in response to actions, which lets a robot predict outcomes and plan; Cosmos and V-JEPA 2 are examples.

Where do I start building?

Try LeRobot or Isaac Lab for learning policies, ROS 2 for robot software and a simulator such as MuJoCo or Gazebo; the hardware section lists edge compute.