Abstract
We introduce BridgeData V2, a large and diverse dataset of roboticmanipulation behaviors designed to facilitate research on scalable robotlearning. BridgeData V2 contains 60,096 trajectories collected across 24environments on a publicly available low-cost robot. BridgeData V2 providesextensive task and environment variability, leading to skills that cangeneralize across environments, domains, and institutions, making the dataset auseful resource for a broad range of researchers. Additionally, the dataset iscompatible with a wide variety of open-vocabulary, multi-task learning methodsconditioned on goal images or natural language instructions. In ourexperiments, we train 6 state-of-the-art imitation learning and offlinereinforcement learning methods on our dataset, and find that they succeed on asuite of tasks requiring varying amounts of generalization. We also demonstratethat the performance of these methods improves with more data and highercapacity models, and that training on a greater variety of skills leads toimproved generalization. By publicly sharing BridgeData V2 and our pre-trainedmodels, we aim to accelerate research in scalable robot learning methods.Project page at https://rail-berkeley.github.io/bridgedata