RLBench
EnvironmentCustom (non-commercial)Introduced 2019-09-26
RLBench is an ambitious large-scale benchmark and learning environment designed to facilitate research in a number of vision-guided manipulation research areas, including: reinforcement learning, imitation learning, multi-task learning, geometric computer vision, and in particular, few-shot learning.
Benchmarks
Robot Manipulation/Succ. Rate (18 tasks, 100 demo/task)Robot Manipulation/Succ. Rate (18 tasks, 10 demo/task)Robot Manipulation/Training Time (V100 x 8 x day)Robot Manipulation/Training Time (A100 x hour)Robot Manipulation/Succ. Rate (10 tasks, 100 demos/task)Robot Manipulation/Succ. Rate (74 tasks, 100 demos/task)Robot Manipulation/Inference Speed (fps)Robot Manipulation/Input Image Size