- Provides pretrained models and datasets usable without assembling physical robots.
- Lowers entry barriers to robotics research with open-source PyTorch tools and community sharing.
- Supports real-world transfer of imitation and reinforcement learning policies.
- Standardized data format simplifies sharing and reproducibility of experiments.