How CodecFlow works
CodecFlow connects the robotics workflow from model to simulation to data to training.
1. Build the robot and scene
Section titled “1. Build the robot and scene”Start in SimArena. Build or import a robot model, then create the environment where you want to test it.
2. Run simulation
Section titled “2. Run simulation”Light simulations can run in the browser. Heavier simulations can move to cloud compute through Fabric.
3. Collect data
Section titled “3. Collect data”Record demonstrations, sensor streams, and scenario outputs from your scene. Export datasets in formats used by robot learning workflows.
4. Train and evaluate
Section titled “4. Train and evaluate”Use the collected data to train policies and evaluate behavior across scenes. Repeat the loop as the robot or task changes.
5. Deploy through the stack
Section titled “5. Deploy through the stack”optr is the runtime layer for robot programs. It is built around See, Think, Act pipelines that connect models, devices, browsers, and cloud compute.