JHU CoSTAR Block Stacking Dataset (BSD)

JHU CoSTAR Block Stacking Dataset (BSD) 是一个由约翰霍普金斯大学计算机科学系创建的数据集，旨在模拟订单履行风格的积木堆叠任务。该数据集包含近12,000次堆叠尝试和超过200万帧的真实数据，适用于研究复杂的、多步骤的机器人操作任务。数据集内容包括动态场景和实时时间序列数据，以及不同颜色的5.1厘米积木。创建过程中，机器人通过与积木互动完成堆叠任务，数据收集利用了CoSTAR系统。该数据集的应用领域广泛，包括但不限于机器人操作、感知研究、深度学习模型的训练与评估，以及解决现实世界中的复杂操作问题。JHU CoSTAR Block Stacking Dataset (BSD) is a dataset developed by the Department of Computer Science at Johns Hopkins University, targeted at simulating order fulfillment-style block stacking tasks. It contains nearly 12,000 stacking attempts and over 2 million frames of real-world data, making it suitable for research on complex, multi-step robotic manipulation tasks. The dataset includes dynamic scenes, real-time time-series data, and 5.1 cm blocks of various colors. During the data collection process, robots completed stacking tasks by interacting with the blocks, and the CoSTAR system was utilized for data acquisition. This dataset has broad application prospects, covering but not limited to robotic manipulation, perception research, training and evaluation of deep learning models, and addressing real-world complex manipulation problems.

搜集汇总

背景与挑战

背景概述

JHU CoSTAR Block Stacking Dataset (BSD) 是由约翰霍普金斯大学创建的数据集，专注于积木堆叠任务，包含近12,000次尝试和超过200万帧真实数据，用于研究复杂的机器人操作。数据集包括动态场景、时间序列数据和5.1厘米彩色积木，通过CoSTAR系统收集，适用于机器人感知、深度学习等领域的训练与评估。