
Fern Robotics Evaluation Platform
Fern learns world models from robot data for scalable policy evaluation and training.
- Currently Offered

Fern learns world models from robot data for scalable policy evaluation and training.
The Fern Robotics Evaluation Platform trains high-fidelity, action-conditioned world models from robot teleoperation data. It generates multi-view RGB rollouts driven by end-effector commands, enabling closed-loop policy evaluation, apples-to-apples comparisons, and large-scale reinforcement learning without standing up a robot fleet.
The platform is offered as a managed cloud service and can also fit custom world models to a customer's own embodiments and historical teleop data.


A learned world model that serves as a virtual simulator for robot policy evaluation and training.


Arc is Saturn Dynamics' world model that generates physical simulations for embodied AI.



Instance verifies robot task completion from camera footage, returning grounded success evidence for every episode.




Realistic simulation environments for training and evaluating robot policies at scale.


OVERPASS is a Game of Drones simulation environment for testing drone, robotics, and AI/ML concepts.
A learned world model that serves as a virtual simulator for robot policy evaluation and training.

Arc is Saturn Dynamics' world model that generates physical simulations for embodied AI.

Horizon is a new physics engine for dexterous manipulation.

Instance verifies robot task completion from camera footage, returning grounded success evidence for every episode.

Isaac Sim is NVIDIA's OpenUSD-based robotics simulation application, built on Omniverse, for developing, testing, and training AI-driven robots.

MuJoCo is a physics engine for robot simulation and reinforcement learning.

Realistic simulation environments for training and evaluating robot policies at scale.

OVERPASS is a Game of Drones simulation environment for testing drone, robotics, and AI/ML concepts.



Robotics Data Infrastructure is XDOF's data pipeline, collection, and annotation platform for robotics foundation models.


Gazebo Sim is an open-source 3D robotics simulator for algorithm development and testing.


Genie 3 is a DeepMind world model that turns text prompts into photorealistic, real-time explorable 3D environments.


World models are AI systems that understand physical reality through continuous observation and interaction rather than processing text alone.



Marble, World Labs' world-model product, generates editable, persistent 3D worlds from text, images, video, or 3D layouts.





Project AirSim is IAMAI's open-source high-fidelity simulator for building and testing autonomous aircraft and robotic systems.


Shotwell Robotics Observability is the observability layer for robotics, turning the video data robots already record into automatic failure detection.


Scale AI — Physical AI powers next-gen robotics with massive, enriched datasets for physical AI breakthroughs.
Robotics Data Infrastructure is XDOF's data pipeline, collection, and annotation platform for robotics foundation models.

Gazebo Sim is an open-source 3D robotics simulator for algorithm development and testing.

Genie 3 is a DeepMind world model that turns text prompts into photorealistic, real-time explorable 3D environments.

World models are AI systems that understand physical reality through continuous observation and interaction rather than processing text alone.

World foundation model platform for generating physically grounded video and training physical AI systems.

Marble, World Labs' world-model product, generates editable, persistent 3D worlds from text, images, video, or 3D layouts.

GWM-1 is Runway's first general world model family for real-time, interactive simulation.

Webots is an open-source robotics simulation software from Cyberbotics.

CoppeliaSim is a robotics simulation software from Coppelia Robotics.

Project AirSim is IAMAI's open-source high-fidelity simulator for building and testing autonomous aircraft and robotic systems.

Shotwell Robotics Observability is the observability layer for robotics, turning the video data robots already record into automatic failure detection.

Scale AI — Physical AI powers next-gen robotics with massive, enriched datasets for physical AI breakthroughs.
