RLBench: The Robot Learning Benchmark & Learning Environment
100 hand-designed manipulation tasks with demos; a staple for 3D policy methods like PerAct.100 个手工设计的操作任务并附演示，是 PerAct 等 3D 策略方法的常用基准。
BibTeX
@misc{james2019rlbench,
title = {{RLBench: The Robot Learning Benchmark \& Learning Environment}},
author = {Stephen James and Zicong Ma and David Rovick Arrojo and Andrew J. Davison},
year = {2019},
eprint = {1909.12271},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/1909.12271},
}