RLBench
100+ manipulation tasks in PyRep. Standard benchmark for VLA and policy evaluation.
Overview
RLBench provides a large set of manipulation tasks in the PyRep (CoppeliaSim) simulation environment. Tasks include pick-place, stacking, opening drawers, and more. Widely used to evaluate vision-language-action models and visuomotor policies.
Key Results (Recent)
- BridgeVLA: 88.2% success
- InternVLA-M1: 95%+ on subsets
Official Links
- github.com/stepjam/RLBench — Code & tasks