Skip to content

Platform Workflow ​

NexCore runs eight stages along a single forward pipeline, with two feedback loops that return to curation. Every stage shares the same model for assets, versions, and permissions.

01Collection taskModel + SN → upload mode → task sheet
02Data assetsFiltered by task · integrity checks
03Curation projectNon-destructive curation · pipeline decision
04Pipeline processingSelect operators · asynchronous run
05Dataset versionHDF5 / LeRobot · frozen on export
06Training and modelsUnified strategy · compute · checkpoints
07DeploymentCloud / edge · NexBench toggle
08Evaluation and feedbackNode scoring · bad case attribution

Stages at a Glance ​

StageInputKey actionsOutputGate
01 Collection taskDevice inventory, scene, target count, collectorSelect device model and SN → select upload mode → enter task detailsTask sheet (format locked automatically, SN bound)Device registration check
02 Data assetsUpload receipts / feedback archivesIntegrity checks, archiving by taskAsset recordsG1 Upload integrity
03 Curation projectSelected asset recordsCreate project; decide whether to run a pipelineCuration project + processing steps—
04 Pipeline processingCuration project with original RAW / HDF5 dataSelect operators, configure parameters, start processingStructured outputs (tagged with pipeline job ID)G2 Pipeline complete
05 Dataset versionCuration results that passed quality checksChoose a format and exportDataset version (frozen)G3 Dataset-ready / G4 Version frozen
06 Training and modelsFrozen dataset versionsSelect model family, apply unified strategy, allocate computeTraining job + model versions / checkpoints—
07 DeploymentModel version, target deviceDeploy; toggle NexBench per jobDeployment record + inference recordG5 Pre-deployment
08 Evaluation and feedbackInference execution data (built-in feedback)Node scoring, bad case attributionEvaluation results + bad casesG6 Feedback review

When something goes wrong at any stage, this table tells you whether an input, an action, or a gate is missing.

Production Loops ​

LoopPathValue
Loop A · Data loopCollect → assets → curation / pipeline → dataset → training → deployment → inference → bad cases → back to curationEach iteration produces a dataset that covers the nodes and scenarios that failed in the last round. Failed samples are never discarded; failure data is ingested too.
Loop B · Version loopDataset frozen on export → training bound to a fixed version → model records its training lineage → deployment records the model version → inference records carry the version → bad cases record their source versionEvery result can answer: which samples were used, what curation was applied, and which version it is pinned to.

NexCore is evolving rapidly; the latest version prevails.NexCore Docs · Embodied AI Skill Learning Platform by Lumos Robotics