---
license: mit
pretty_name: Memory-T-Bench
tags:
  - robotics
  - imitation-learning
  - diffusion-policy
  - memory
  - pomdp
  - push-t
size_categories:
  - 100K<n<1M
task_categories:
  - robotics
---

# Memory-T-Bench

Partially observable, contact-rich 2D pushing tasks derived from
[Push-T](https://diffusion-policy.cs.columbia.edu/), introduced in
**"Remember what you did?: Learning Behavioral Memories for Partially Observable
Object Manipulation"** ([paper](https://arxiv.org/abs/2606.21188) · [project page](https://robo-camp.github.io/)). The
current observation never reveals what to do next; the policy must remember
what it already did or found out.

The release ships **only the modalities a real-world policy has access to**:
image, robot state and action. No privileged hidden-state arrays are included,
so a model cannot read the variable it is supposed to remember.

## Tasks

<table align="center">
<tr>
<td align="center"><img src="https://huggingface.co/datasets/harrywang01/Memory-T-Bench/resolve/main/multi-goals/preview.gif" width="200"><br><b>Multi-Goals</b></td>
<td align="center"><img src="https://huggingface.co/datasets/harrywang01/Memory-T-Bench/resolve/main/swap-direct/preview.gif" width="200"><br><b>Swap-Direct</b></td>
<td align="center"><img src="https://huggingface.co/datasets/harrywang01/Memory-T-Bench/resolve/main/swap-shuffle/preview.gif" width="200"><br><b>Swap-Shuffle</b></td>
<td align="center"><img src="https://huggingface.co/datasets/harrywang01/Memory-T-Bench/resolve/main/find-track/preview.gif" width="200"><br><b>Find-Track</b></td>
</tr>
</table>

| folder | task | what must be remembered | success criterion | episodes | steps |
|---|---|---|---|---:|---:|
| [`multi-goals`](multi-goals/) | **Multi-Goals** — Push the gray T block onto each of the three target areas, each of them only once, in any order. | which targets are already done | each goal reached at ≥ 85 % block–goal overlap, none twice | 320 | 94,636 |
| [`swap-direct`](swap-direct/) | **Swap-Direct** — Push the blue and red T blocks to the middle, stop on the black dot, then move each block into the other's starting area. | where each T block started | both blocks ≥ 85 % overlap with the other's start area | 320 | 105,615 |
| [`swap-shuffle`](swap-shuffle/) | **Swap-Shuffle** — Use the empty target area as a buffer to shuffle the blue and red T blocks into each other's starting area. | where each T block started | both blocks ≥ 85 % overlap with the other's start area | 320 | 118,629 |
| [`find-track`](find-track/) | **Find-Track** — Push the gray T block from the starting area through one of three tracks to the target area on the opposite side; two tracks are impassable. | which track is passable | ≥ 85 % overlap with the target area, no re-tried track | 320 | 91,807 |

Each folder holds `data.zarr.zip` and a `preview.gif`. Environments, data
collection scripts and the CAMP training code are in the
[CAMP repository](https://robo-camp.github.io/) (`memory_t_bench` package).

## Quick start

```python
from huggingface_hub import hf_hub_download
import zarr

path = hf_hub_download(repo_id="harrywang01/Memory-T-Bench",
                       filename="multi-goals/data.zarr.zip", repo_type="dataset")
root = zarr.open(zarr.ZipStore(path, mode="r"), mode="r")
print(root.tree())
ends = root["meta/episode_ends"][:]        # exclusive end index of every episode
ep0 = slice(0, int(ends[0]))
img, state, action = root["data/img"][ep0], root["data/state"][ep0], root["data/action"][ep0]
```

## Data format

Each task is a single `data.zarr.zip` in the Diffusion Policy replay-buffer
convention:

```
data.zarr.zip
├── data/
│   ├── action                (N, 2)             float32   target agent position (pixels)
│   ├── img                   (N, 96, 96, 3)     uint8     rendered observation
│   ├── state                 (N, 2)             float32   agent position (pixels)
│   └── language_instruction  (N,)               <U200     constant per task
└── meta/
    └── episode_ends          (E,)               int64     exclusive end indices
```

Arrays are flat over all episodes; episode `i` spans
`[episode_ends[i-1], episode_ends[i])`.

## Citation

```bibtex
@misc{wang2026rememberdidlearningbehavioral,
      title={Remember what you did?: Learning Behavioral Memories for Partially Observable Object Manipulation},
      author={Kuancheng Wang and Seungho Yeom and Jinglin Cao and Yuheng Zhi and Nikhil Shinde and Michael Yip},
      year={2026},
      eprint={2606.21188},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2606.21188},
}
```

## License

MIT — see [`LICENSE`](LICENSE).