/ INTEGRATIONS

Fits the stack you already run.

Every dataset lands in the tools your team opens every day — training frameworks, simulators, visualizers, and experiment trackers, with no glue code to maintain.

8integrations
4topics

/ Middleware and tooling

ROS 2

Record robot demonstrations from a live ROS 2 graph with rosbag2, avoid the QoS and clock pitfalls that corrupt recordings, and export to training formats.

6 min read

Foxglove

Use Foxglove to visually inspect and QA robot demonstration recordings — live or from MCAP — with 3D, image, and plot panels built for robotics data.

5 min read

/ Training and datasets

LeRobot

LeRobot is Hugging Face's robotics library. Here is how a Telemanual export maps onto LeRobotDataset, and how to load, train, and evaluate against it.

5 min read

PyTorch

Turning a captured robot dataset into a PyTorch pipeline — Dataset choices, video decode bottlenecks, episode-safe splits, and reproducible sampling.

5 min read

Hugging Face Hub

Host robot datasets on the Hugging Face Hub with dataset cards, access control, and pinned revisions — the layer between raw capture and a citable release.

5 min read

/ Simulation

NVIDIA Isaac Lab

How NVIDIA Isaac Lab fits a real-data workflow — digital twins, HDF5 demonstration replay, sim-and-real co-training, and the limits of sim evaluation.

5 min read

MuJoCo

MuJoCo is DeepMind's open-source physics engine for contact-rich robotics research. Here is how recorded demonstration data connects to MJCF models.

5 min read

/ Experiment tracking

Weights & Biases

Tracking robot-learning runs in Weights & Biases — versioning datasets as artifacts, logging rollout videos, and why validation loss alone falls short.

7 min read

Put this into practice.

Tell us what your robots need to learn. We will scope the rig, the operators, the protocol, and the first datasets — usually in one call.

Start a pilotRead the docs