CERTIS is a collection of high-quality robotic surgery environments and tasks for NVIDIA Isaac Lab and NVIDIA Isaac Sim. It brings reusable surgical simulation work into one project while acting as a feeder for the upstream projects on which it builds.
CERTIS aims to provide a reference collection of environments representing the most frequent tasks in robotic surgery. Each environment will support teleoperation, model training and reproducible evaluation.
CERTIS is intended to serve two complementary purposes:
- to act as a set of gold-standard benchmark environments and tasks, or gyms, for surgical robotics
- and to provide a library of compositional primitives from which more complex surgical tasks and workflows can be assembled.
- Isaac-native: environments are implemented in Isaac Lab and Isaac Sim using shared conventions and reusable components.
- Teleoperation-ready: tasks are designed for demonstration collection, human control and embodied policy development.
- Benchmark-quality: environments provide clear task definitions, success and failure conditions, metrics and reproducible evaluation protocols.
- Composable: robots, scenes, assets, task logic and interaction primitives can be combined without rebuilding complete environments from scratch.
- Visually excellent: environments use high-quality assets, materials, lighting and rendering suitable for both model training and compelling visualisation.
- Upstream-oriented: mature environments and general-purpose components are prepared for contribution to relevant upstream projects.
CERTIS is for researchers and engineers working on surgical robotics, teleoperation, imitation learning, reinforcement learning, vision-language-action models, simulation and embodied evaluation.