SERIES / TECHNICAL NOTES
Inside Robotic Systems
A series on the stack behind modern robots — middleware, control, perception, simulation, and learning.
Inside ROS 2: Anatomy of a Robot Communication System
An under-the-hood tour of ROS 2’s communication fabric—from DDS and core primitives to QoS, executors, and zero-copy paths for large model payloads.
Inside Robot Data Storage: MCAP, Zarr, and Multimodal Datasets at Scale
How production robot fleets store multimodal data—MCAP and Zarr layouts, sharding, featurization, and the dominant production patterns (plus academic alternatives) for datasets that stay queryable at scale.
Inside Robot Data Pipelines: Filtering, Autolabeling, and Training Processors
How robot datasets become training-ready—filtering and curation, dataset mixing and action-space harmonization, autolabeling and dataset enrichment, online training-time processors, and the distributed infrastructure that runs all of it at fleet scale.
Subscribe
Occasional notes when a new piece in this series goes up. No extras.