Robotics and autonomous vehicle teams can lose a week to setup before their first run. CoreWeave Physical AI Launchables replaces that week with a catalog of workloads that CoreWeave engineers build and test before you run them. In this brief, learn how to:
- Launch evaluation, synthetic data generation, and policy training Templates on NVIDIA Isaac Sim, CARLA, MuJoCo, NVIDIA AlpaSim, and NVIDIA Cosmos 3 with one CLI
- Change the settings that matter with name=value pairs, or start from a blank Template for workloads that aren’t in the catalog
- Check what each run produced and what it missed, through an interface that scripts and agents can drive the same way you do
