The CoSTAR Block Stacking Dataset contains nearly 12,000 recordings of a robot attempting to stack colored blocks under variable conditions. The dataset includes calibrated RGB-D images, joint angles, and joint velocities with timestamps. There are a total of nearly 12,000 stacking attempts and over 2 million frames.
The dataset is designed as a benchmark for performing complex, multi-step manipulation tasks in challenging scenes and may be useful in the development of models for grasping and stacking.
Verified
@article{hundt2019costar,
author = {Andrew Hundt and Varun Jain and Chia-Hung Lin and Chris Paxton and Gregory D. Hager},
journal = {Intelligent Robots and Systems (IROS), 2019 IEEE International Conference on},
title = {The CoSTAR Block Stacking Dataset: Learning with Workspace Constraints},
url = {https://arxiv.org/abs/1810.11714},
year = {2019}
}
