All Tools
R
AgentsFreeOpen Source
RLBENCH
Large-scale robot manipulation benchmark and learning env
ABOUT
Vision-guided manipulation research needs many tasks, demos, and few-shot splits, not a single reach-target env. RLBench puts 100-plus CoppeliaSim tasks behind one API with demonstrations and Gym wrappers so imitation and multi-task agents share a benchmark.
INSTALL
pip install git+https://github.com/stepjam/RLBench.gitINTEGRATION GUIDE
1. Train few-shot manipulation agents on FS10 or FS25 task splits
2. Collect demonstration trajectories for imitation learning
3. Run Gym-compatible RL on reach, pick, and open-door tasks
4. Domain-randomize visuals for sim-to-real robot transfer
TAGS
pythonagentsroboticsreinforcement-learningimitation-learningsimulationgym