HuMeGo index page
Browse files
README.md
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-4.0
|
| 3 |
+
task_categories:
|
| 4 |
+
- robotics
|
| 5 |
+
tags:
|
| 6 |
+
- LeRobot
|
| 7 |
+
- HuMeGo
|
| 8 |
+
- humanoid
|
| 9 |
+
- unitree-g1
|
| 10 |
+
- memory
|
| 11 |
+
- loco-manipulation
|
| 12 |
+
- benchmark
|
| 13 |
+
pretty_name: "HuMeGo: humanoid memory benchmark (Memory before Goal)"
|
| 14 |
+
---
|
| 15 |
+
|
| 16 |
+
# HuMeGo: Humanoid Memory Benchmark
|
| 17 |
+
|
| 18 |
+
**HuMeGo** is the memory-dependent humanoid loco-manipulation benchmark of
|
| 19 |
+
*Memory before Goal: Role of Pre-History Memory for Humanoid Loco-Manipulation*
|
| 20 |
+
([arXiv 2610.31166](https://arxiv.org/abs/2610.31166) · [GitHub](https://github.com/crystalccy1/MemoryBeforeGoal) ·
|
| 21 |
+
[project page](https://memory-before-goal.github.io) · [collection](https://huggingface.co/collections/ChenyuanC/humego-6ada5cbea74d4cc678c91c68)).
|
| 22 |
+
|
| 23 |
+
In every task the correct action depends on a cue that has already disappeared and is not named in the instruction:
|
| 24 |
+
the robot must remember what happened before it acts on its goal. HuMeGo has six simulated tasks and six real-world
|
| 25 |
+
tasks on a Unitree G1 humanoid, covering object, spatial and temporal memory. This page is an index; the data are in
|
| 26 |
+
the repositories below, all in [LeRobot](https://github.com/huggingface/lerobot) v2.1 format under CC BY 4.0.
|
| 27 |
+
|
| 28 |
+
## Simulation (Unitree G1, 50 Hz, head camera)
|
| 29 |
+
|
| 30 |
+
100 scripted expert demonstrations per task for training, and 90 test scenes per task (30 per randomization level L0 / L1 / L2).
|
| 31 |
+
|
| 32 |
+
| Task | Memory | Cue | Training | Test scenes |
|
| 33 |
+
|---|---|---|---|---|
|
| 34 |
+
| FlashPad | counting | a lamp flashes one to three times | [HuMeGo-Sim-FlashPad](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Sim-FlashPad) | [TestScenes/FlashPad](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Sim-TestScenes/tree/main/FlashPad) |
|
| 35 |
+
| HideMug | permanence | a mug hides behind one of three screens | [HuMeGo-Sim-HideMug](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Sim-HideMug) | [TestScenes/HideMug](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Sim-TestScenes/tree/main/HideMug) |
|
| 36 |
+
| ShellGame | permanence | cups are swapped over a hidden ball | [HuMeGo-Sim-ShellGame](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Sim-ShellGame) | [TestScenes/ShellGame](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Sim-TestScenes/tree/main/ShellGame) |
|
| 37 |
+
| LitLamp | reference | one lamp lights up for 0.5 s | [HuMeGo-Sim-LitLamp](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Sim-LitLamp) | [TestScenes/LitLamp](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Sim-TestScenes/tree/main/LitLamp) |
|
| 38 |
+
| BlinkPad | reference | one pad blinks for 0.2 s | [HuMeGo-Sim-BlinkPad](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Sim-BlinkPad) | [TestScenes/BlinkPad](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Sim-TestScenes/tree/main/BlinkPad) |
|
| 39 |
+
| TouchPad | reference | one pad blinks, then the robot turns away | [HuMeGo-Sim-TouchPad](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Sim-TouchPad) | [TestScenes/TouchPad](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Sim-TestScenes/tree/main/TouchPad) |
|
| 40 |
+
|
| 41 |
+
## Real world (Unitree G1, whole-body teleoperation with SONIC, 30 fps head camera)
|
| 42 |
+
|
| 43 |
+
15 teleoperated demonstrations per task, five per answer.
|
| 44 |
+
|
| 45 |
+
| Task | Memory | Dataset |
|
| 46 |
+
|---|---|---|
|
| 47 |
+
| Trash bag | where | [HuMeGo-Real-TrashBag](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Real-TrashBag) |
|
| 48 |
+
| Remember-chip | what | [HuMeGo-Real-RememberChip](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Real-RememberChip) |
|
| 49 |
+
| Café A: first come, first served | when | [HuMeGo-Real-CafeA](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Real-CafeA) |
|
| 50 |
+
| Café B: which seat has the cola | what + where | [HuMeGo-Real-CafeB](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Real-CafeB) |
|
| 51 |
+
| Café C: where did the cola stop first | what + where + when | [HuMeGo-Real-CafeC](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Real-CafeC) |
|
| 52 |
+
| Chip-cart | progress | [HuMeGo-Real-ChipCart](https://huggingface.co/datasets/ChenyuanC/HuMeGo-Real-ChipCart) |
|
| 53 |
+
|
| 54 |
+
## Download
|
| 55 |
+
|
| 56 |
+
```bash
|
| 57 |
+
huggingface-cli download ChenyuanC/HuMeGo-Sim-FlashPad --repo-type dataset --local-dir HuMeGo-Sim-FlashPad
|
| 58 |
+
```
|
| 59 |
+
|
| 60 |
+
## License
|
| 61 |
+
|
| 62 |
+
Released under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/).
|
| 63 |
+
|
| 64 |
+
## Citation
|
| 65 |
+
|
| 66 |
+
```bibtex
|
| 67 |
+
@misc{chen2026memorybeforegoal,
|
| 68 |
+
title={Memory before Goal: Role of Pre-History Memory for Humanoid Loco-Manipulation},
|
| 69 |
+
author={Chenyuan Chen and Ran Ding and Lechen Zhang and Zewen He and Yoshihiko Nakamura},
|
| 70 |
+
year={2026},
|
| 71 |
+
eprint={2610.31166},
|
| 72 |
+
archivePrefix={arXiv},
|
| 73 |
+
primaryClass={cs.RO},
|
| 74 |
+
url={https://arxiv.org/abs/2610.31166},
|
| 75 |
+
}
|
| 76 |
+
```
|