README.md

---
license: apache-2.0
task_categories:
- robotics
library_name: robotics
---

# MemoryBench Dataset

MemoryBench is a benchmark dataset designed to evaluate spatial memory and action recall in robotic manipulation. This dataset accompanies the **SAM2Act+** framework, introduced in the paper *[SAM2Act: Integrating Visual Foundation Model with A Memory Architecture for Robotic Manipulation](https://huggingface.co/papers/2501.18564)*. For detailed task descriptions and more information about this paper, please visit SAM2Act's [website](https://sam2act.github.io). Code can be found at [https://github.com/sam2act/sam2act](https://github.com/sam2act/sam2act).

The dataset contains scripted demonstrations for three memory-dependent tasks designed in RLBench (same version as the one used in [PerAct](https://peract.github.io/)):

- **Reopen Drawer**: Tests 3D spatial memory along the z-axis.
- **Put Block Back**: Evaluates 2D spatial memory along the x-y plane.
- **Rearrange Block**: Requires backward reasoning based on prior actions.

## Dataset Structure

The dataset is organized as follows:
```
data/
├── train/  # 100 episodes per task
├── test/   # 25 episodes per task
└── files/  # task files (.ttm & .py)
```

- **data/train/**: Contains three zip files, each corresponding to one of the three tasks. Each zip file contains **100** scripted demonstrations for training.
- **data/test/**: Contains the same three zip files, but each contains **25** held-out demonstrations for evaluation.
- **data/files/**: Includes necessary `.ttm` and `.py` files for running evaluation.

## Usage

This dataset is designed for use in the same manner as the RLBench 18 Tasks proposed by [PerAct](https://peract.github.io/). You can follow the same usage guidelines or stay updated with SAM2Act's [code repository](https://github.com/sam2act/sam2act) for further instructions.

## Acknowledgement

We would like to acknowledge [Haoquan Fang](https://hq-fang.github.io/) for leading the conceptualization of MemoryBench, providing key ideas and instructions for task design, and [Wilbert Pumacay](https://wpumacay.github.io/) for implementing the tasks and ensuring their seamless integration into the dataset. Their combined efforts, along with the oversight of [Jiafei Duan](https://duanjiafei.com/) and all co-authors, were essential in developing this benchmark for evaluating spatial memory in robotic manipulation.

## Citation

If you use this dataset, please cite the SAM2Act paper:

```bibtex
@misc{fang2025sam2act,
      title={SAM2Act: Integrating Visual Foundation Model with A Memory Architecture for Robotic Manipulation},
      author={Haoquan Fang and Markus Grotz and Wilbert Pumacay and Yi Ru Wang and Dieter Fox and Ranjay Krishna and Jiafei Duan},
      year={2025},
      eprint={2501.18564},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2501.18564},
}
```