# Wano Robotics — the studio for physical AI Wano Studio brings robotics development into one local-first workspace: design a scene, choose a robot, define its task, review demonstrations and prepare training. Wano Framework connects that workflow through a declarative blueprint, with simulation targets and ROS 2 deployment tooling. Spend less effort connecting separate tools and more effort developing the skill your robot needs. ## Why choose Wano - Keep control of your work: the local-first workflow keeps scenes, datasets and model weights on your own machine. - Build from one source: a declarative blueprint connects the scene, robot and task instead of requiring separate descriptions for every tool. - Work with your robotics ecosystem: Isaac Lab, MuJoCo, Gazebo and Webots are simulation targets, alongside browser physics with MuJoCo WASM and Rapier. - Keep your outputs usable: generated code and standard dataset formats reduce dependence on a single application. - Bring AI agents into the workflow: MCP tools expose studio operations, with a Guard Engine validating actions before they are applied. ## Availability Early access is planned for Q4 2026, in small cohorts. The roadmap marks the web studio, 3D scene editor, Isaac Sim streaming and Wano Framework as shipped. Multi-simulator end-to-end workflows, dataset tooling and GPU fine-tuning are in progress. The eight-chapter photo/video journey illustrates the product vision; it is not a claim that every illustrated capability is available today. Managed cloud GPUs and VR teleoperation are planned. ## Inside Wano Studio ### Build the environment in a 3D editor Drop a table, props, lights and cameras into the scene, place them with the gizmo, and set the task zones. Everything you place is written into the blueprint — the single declarative source the studio compiles from. [Explore this step](https://wanorobotics.com/#product-compose) ### Pick from 150 models, or bring your own Arms, mobile bases, quadrupeds, humanoids, hands and drones ship with the studio, each with its actuators, sensors and meshes already resolved. Import your own from a URDF or MJCF file when the catalog doesn’t have it. [Explore this step](https://wanorobotics.com/#product-robot) ### Frame the mission in a few answers Choose the behavior, robot and environment while the live recap stays in view. Wano checks the composition as you go, then seeds the complete mission — task zones, success condition and training branch included. [Explore this step](https://wanorobotics.com/#product-define) ### Take control and record the behavior Drive the robot from one cockpit while the 3D workcell, controls, telemetry and latest takes stay visible together. Every accepted motion becomes a synchronized episode ready for review. [Explore this step](https://wanorobotics.com/#product-demonstrate) ### Replay every episode before it becomes data One timeline drives the 3D replay, task phases and synchronized signals. Review the motion, inspect the active steps and accept or reject the take with the evidence still on screen. [Explore this step](https://wanorobotics.com/#product-curate) ### Watch every run converge Launch a training run from the curated data, then follow convergence and compare every run. Progress, checkpoints and GPU state stay beside the metrics, so the run is readable without leaving the studio. [Explore this step](https://wanorobotics.com/#product-train) ## Eight chapters — a vision for robot learning The photo/video workflows below illustrate the product vision, rather than a list of generally available features. ### Photo to 3D scene Take a photograph of a desk. AI identifies the objects and their arrangement, then rebuilds the workcell using editable 3D assets from the library. Refine the geometry, physics and task zones in the studio. A table stays a table. The cube, blue cylinder, tray and surrounding space become objects the simulator can work with. Input: Desk photograph. Output: Editable 3D blueprint. [Explore this chapter](https://wanorobotics.com/#photo-to-scene) ### Synthetic teleoperation A video captures a hand picking up an object and placing it in a tray. Identify the hand, the manipulated object and the action over time. Reconstruct the scene, then retarget the gesture to the robot in simulation. Synthetic teleoperation: demonstrations generated from video, without manually driving the robot for every take. Input: Human demonstration video. Output: Robot motion in simulation. [Explore this chapter](https://wanorobotics.com/#synthetic-teleoperation) ### One scene. Many variations. One demonstration branches into a family of variations. Change the cube’s colour, the tabletop, the wall, the object positions or the motion path. The goal stays the same: put the object in the tray. Appearance, placement and movement vary together. Every branch adds another way to learn the same skill. Input: One reconstructed demonstration. Output: Task-preserving variants. [Explore this chapter](https://wanorobotics.com/#scene-variants) ### Parallel simulation Send the scene and its variants to the GPU. Parallel virtual environments replay the behaviour under different conditions. Each environment produces its own observations, actions and outcomes. The workstation authors the scene. The GPU scales the experience. More environments mean more opportunities to collect useful episodes. Input: Blueprints + GPU compute. Output: Parallel simulation episodes. [Explore this chapter](https://wanorobotics.com/#parallel-simulation) ### Synthetic datasets Every accepted episode becomes a synchronized sequence: what the robot saw, what it did, and the instruction it followed. Review the trajectories and keep useful examples before exporting the dataset. Images, actions and language stay aligned. Export in the formats your training stack expects, including LeRobot, HDF5 and RLDS. Input: Simulation recordings. Output: Curated synthetic dataset. [Explore this chapter](https://wanorobotics.com/#synthetic-datasets) ### Train the VLA Bring a pretrained vision-language-action model and the curated dataset together on the GPU. Fine-tuning adapts the model to the task, its objects and the robot’s actions. Follow the run and keep its checkpoints. Vision describes what the robot sees. Language describes the goal. Action is what the robot does next. Input: Dataset + pretrained VLA. Output: Task-adapted model weights. [Explore this chapter](https://wanorobotics.com/#vla-training) ### Validate the skill Run the trained policy in held-out environments. Change colours, positions and trajectories again, then inspect whether the robot still completes the task. Use the failures to decide what to record or train next. A policy should handle new conditions, not simply replay a familiar frame. Validation closes the loop before deployment. Input: Adapted policy + unseen scenes. Output: Evaluated robot behaviour. [Explore this chapter](https://wanorobotics.com/#policy-validation) ### Deploy to the real world Export the evaluated policy and connect it to the real robot through the deployment runtime. The robot observes its surroundings, receives the task instruction and predicts the actions that move the object into the tray. The learned skill runs on the robot: camera images and a task instruction go in, joint actions come out. The robot repeats this loop as it moves. Input: Validated skill + task instruction. Output: Real-world robot action. [Explore this chapter](https://wanorobotics.com/#real-world-deployment) ## Frequently asked questions ### Do I need Isaac Sim to use it? No. MuJoCo, Gazebo and Webots are first-class targets, and two physics engines run directly in the browser (MuJoCo WASM, Rapier). Isaac Lab is one of six targets — the one you’ll want for GPU-parallel RL. ### Does my data leave my machine? No. Scenes, datasets, model weights and training runs stay local. A SaaS tier is planned for teams that want managed GPUs — it will be opt-in, not the default. ### Which robots are supported? 150 models ship in the catalog — arms (SO-ARM100, ALOHA 2), quadrupeds (Spot), humanoids and more. Anything with a URDF or MJCF can be imported. ### Is the generated code mine? Yes. Generated Python, ROS 2 packages and datasets are plain files on your disk, in standard formats (LeRobot, HDF5, RLDS). No lock-in by construction: if you stop using Wano, the code still runs. ### When does access open? Q4 2026, in small cohorts. ROS 2 developers, Isaac Lab users and LeRobot teams get priority. Register below and we’ll be in touch. ## Build the next skill with Wano Bring your robot, keep control of your data, and connect the steps that turn an idea into a robot-learning workflow. [Join the early-access waitlist](https://wanorobotics.com/#access).