DROID: A Large-Scale In-The-Wild Robot Manipulation Dataset

By Alexander Khazatsky, Karl Pertsch, S. Nair, Ashwin Balakrishna, Sudeep Dasari, Siddharth Karamcheti, Soroush Nasiriany, M. K. Srirama, L. Chen, Kirsty Ellis and 90 more. Robotics: Science and Systems Conference, 2024

905 citations recorded by the source corpus.

The creation of large, diverse, high-quality robot manipulation datasets is an important stepping stone on the path toward more capable and robust robotic manipulation policies. However, creating such datasets is challenging: collecting robot manipulation data in diverse environments poses logistical and safety challenges and requires substantial investments in hardware and human labour. As a result, even the most general robot manipulation policies today are mostly trained on data collected in a small number of environments with limited scene and task diversity. In this work, we introduce DROID (Distributed Robot Interaction Dataset), a diverse robot manipulation dataset with 76k demonstration trajectories or 350 hours of interaction data, collected across 564 scenes and 84 tasks by 50 data collectors in North America, Asia, and Europe over the course of 12 months. We demonstrate that training with DROID leads to policies with higher performance and improved generalization ability. We open source the full dataset, policy learning code, and a detailed guide for reproducing our robot hardware setup.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Control and Systems Engineering

View Paper: https://www.semanticscholar.org/paper/ee5070fe52fd17da9a89d3f342fb07cc9ae51afe

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Control and Systems Engineering

Citations in the corpus

Evaluating Real-World Robot Manipulation Policies in Simulation (2024, positive): “Much recent work in robotic manipulation has demonstrated impressive tasks that go beyond rigid object manipulation [20, 10, 33].”
AgiBot World Colosseo: A Large-Scale Manipulation Platform for Scalable and Intelligent Embodied Systems (2025, positive): “GO-1 exhibits exceptional capability in real-world dexterous and long-horizon tasks, achieving over 60% success rate on complex tasks and outperforming prior RDT approach by 32%.”
SpatialVLA: Exploring Spatial Representations for Visual-Language-Action Model (2025, neutral): “Moreover, as discussed in OpenVLA [30], DROID dataset [29] are removed from the data mixture for the final third of pre-training to improve the quality of the pre-trained SpatialVLA model.”
UniVLA: Learning to Act Anywhere with Task-centric Latent Actions (2025, positive): “Empowered by the emergence of large-scale robotic datasets [78, 63, 38, 1], robot policies based on vision-language-action models (VLA) have made encouraging strides recently [10, 28, 39].”
OmniH2O: Universal and Dexterous Human-to-Humanoid Whole-Body Teleoperation and Learning (2024, background): “Recent efforts [40, 41, 42, 43, 44, 45, 46, 47] have focused on collecting robotic data using various embodiments for different tasks.”
Deep Reinforcement Learning for Robotics: A Survey of Real-World Successes (2024, background): “Lastly, recent advances in large-scale robot dataset (205, 206) and robot foundation models (207, 208) present exciting open opportunities for RL successes in the real world.”
Open-TeleVision: Teleoperation with Immersive Active Visual Feedback (2024, neutral): “If we distinguish the existing work with sources of demonstrations, we can classify them into learning from real-robot expert data [14, 10, 12, 59, 60, 61, 62, 50, 63, 1, 64, 51, 65], learning from play data [66, 67, 68], and learning from human demonstrations [67, 69, 70, 71, 72, 73, 74, 56].”
Robotic Control via Embodied Chain-of-Thought Reasoning (2024, neutral): “Towards this goal, recent works have explored training “generalist robot policies” [10, 11, 12, 13, 14, 15, 16, 17] on diverse robot datasets [18, 19, 10, 20, 21, 22, 23, 24, 13, 14, 25, 15, 26, 6].”
A Survey on Vision--Language--Action Models for Embodied AI (2024, neutral): “Table adapted from [146], [180]. based on how easily they can be achieved, a crucial aspect often overlooked by other high-level task planners.”
OpenVLA: An Open-Source Vision-Language-Action Model (2024, methodology): “Generalist Robot Policies A recent trend in robotics works towards training multi-task “generalist” robot policies [2, 6, 45–49] on large diverse robot datasets [1, 2, 6, 11, 45, 49–56], spanning many different robot embodiments [1, 5, 53, 57–66].”