Robo-DM: Data Management For Large Robot Datasets

TL;DR

Robo-DM is an open-source cloud-based toolkit that efficiently manages, compresses, and accelerates loading of large robot datasets, enabling scalable training without sacrificing task performance.

Contribution

It introduces Robo-DM, a novel data management system that significantly reduces dataset size and improves data loading speed for large robot datasets.

Findings

Robo-DM achieves up to 70x data compression (lossy) and 3.5x (lossless).

Robo-DM accelerates data retrieval by up to 50x compared to similar frameworks.

Model training with Robo-DM's compressed data maintains accuracy in robot tasks.

Abstract

Recent results suggest that very large datasets of teleoperated robot demonstrations can be used to train transformer-based models that have the potential to generalize to new scenes, robots, and tasks. However, curating, distributing, and loading large datasets of robot trajectories, which typically consist of video, textual, and numerical modalities - including streams from multiple cameras - remains challenging. We propose Robo-DM, an efficient open-source cloud-based data management toolkit for collecting, sharing, and learning with robot data. With Robo-DM, robot datasets are stored in a self-contained format with Extensible Binary Meta Language (EBML). Robo-DM can significantly reduce the size of robot trajectory data, transfer costs, and data load time during training. Compared to the RLDS format used in OXE datasets, Robo-DM's compression saves space by up to 70x (lossy) and 3.5x (lossless). Robo-DM also accelerates data retrieval by load-balancing video decoding with memory-mapped decoding caches. Compared to LeRobot, a framework that also uses lossy video compression, Robo-DM is up to 50x faster when decoding sequentially. We physically evaluate a model trained by Robo-DM with lossy compression, a pick-and-place task, and In-Context Robot Transformer. Robo-DM uses 75x compression of the original dataset and does not suffer reduction in downstream task accuracy.