Intake
Upload multi-modal robotics raw data
Drag and drop MCAP, ROSBag2, video, and CSV files, or load a sample to get started.
Product platform
BYOD Processing Platform
Bring Your Own Data for robotics teams preparing multi-modal AI training data. Upload raw files, prepare dataset structure, convert to LeRobot-compatible artifacts, validate quality, and export model-ready outputs.
Bring Your Own Data Start with your own robotics logs, videos, CSV files, and project context
Multi-Modal Intake Prepare raw robotics data through a guided upload and validation flow
LeRobot Output Generate compatible artifacts for downstream model training
Quality Review Inspect validation signals before exporting model-ready datasets
Workflow
From raw robotics files to model-ready data in simple steps.
BYOD reduces the manual preparation work between raw robot recordings and usable AI training data. Teams can upload multi-modal source files, prepare the dataset structure, run conversion jobs, and review quality before export.
Configure
Configure and map inputs
Choose the robot, timing, and LeRobot version, then auto-detect or confirm topic mappings.
Process
Run the guided conversion
BYOD inspects, decodes, aligns, and writes the inputs into private LeRobot artifacts.
Validate
Review quality and use the dataset
Inspect job metrics and validation results, then access model-ready output from Datasets.
Next step
Prepare robotics data for training with less manual setup.
Use BYOD to move from multi-modal robotics raw data to validated, model-ready datasets with a guided upload, conversion, and quality review flow.