Update README.md
Browse files
README.md
CHANGED
|
@@ -24,8 +24,14 @@ size_categories:
|
|
| 24 |
>
|
| 25 |
> IEEE Robotics and Automation Letters (RAL), 2025
|
| 26 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
**Mimir** is a hierarchical goal-driven diffusion model for end-to-end autonomous driving. It improves upon **GoalFlow** by explicitly modeling goal uncertainty and accelerating goal inference, enabling more robust and efficient planning. With a ResNet-34 backbone, Mimir achieves 89.3 PDMS and 34.6 EPDMS, demonstrating strong performance in both accuracy and efficiency.
|
| 28 |
|
|
|
|
|
|
|
| 29 |
This repository contains the **model weights and intermediate data files** required for Mimir's evaluation and training. For the full codebase, please visit the [main Mimir repository](https://github.com/ZebinX/Mimir-Uncertainty-Driving).
|
| 30 |
|
| 31 |
---
|
|
@@ -45,29 +51,15 @@ This repository contains the **model weights and intermediate data files** requi
|
|
| 45 |
| `frame_mapping_navhard.yaml` | Frame index mapping for Navhard |
|
| 46 |
| `global_pose_navhard.npy` | Global pose data for Navhard |
|
| 47 |
|
| 48 |
-
> Each `*_naviunc/` directory contains `navi.npy` (navigation points) and `unc.npy` (uncertainty estimates).
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
| 52 |
## 📊 Results
|
| 53 |
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|:---|:---:|:---:|:---:|:---:|:---:|
|
| 58 |
-
| GoalFlow‡ | 98.2 | 96.4 | 93.8 | 82.6 | 87.9 |
|
| 59 |
-
| DiffusionDrive | 98.2 | 96.2 | 94.7 | 82.2 | 88.1 |
|
| 60 |
-
| **Mimir** | **98.2** | **97.5** | 94.6 | **83.6** | **89.3** |
|
| 61 |
-
|
| 62 |
-
### Navhard (NAVSIM v2)
|
| 63 |
|
| 64 |
-
|
| 65 |
-
|:---|:---:|
|
| 66 |
-
| GoalFlow† | 28.9 |
|
| 67 |
-
| DiffusionDrive† | 29.1 |
|
| 68 |
-
| **Mimir** | **34.6** |
|
| 69 |
|
| 70 |
-
Please refer to the [paper](https://arxiv.org/pdf/2512.07130) for
|
| 71 |
|
| 72 |
---
|
| 73 |
|
|
@@ -90,4 +82,4 @@ If you find Mimir useful, please consider citing our paper:
|
|
| 90 |
|
| 91 |
## Contact
|
| 92 |
|
| 93 |
-
If you have any questions or suggestions, please open an issue or contact: xzebin@bupt.edu.cn
|
|
|
|
| 24 |
>
|
| 25 |
> IEEE Robotics and Automation Letters (RAL), 2025
|
| 26 |
|
| 27 |
+
<div align="center">
|
| 28 |
+
<img src="./assets/teaser.png" width="75%" />
|
| 29 |
+
</div>
|
| 30 |
+
|
| 31 |
**Mimir** is a hierarchical goal-driven diffusion model for end-to-end autonomous driving. It improves upon **GoalFlow** by explicitly modeling goal uncertainty and accelerating goal inference, enabling more robust and efficient planning. With a ResNet-34 backbone, Mimir achieves 89.3 PDMS and 34.6 EPDMS, demonstrating strong performance in both accuracy and efficiency.
|
| 32 |
|
| 33 |
+
|
| 34 |
+
|
| 35 |
This repository contains the **model weights and intermediate data files** required for Mimir's evaluation and training. For the full codebase, please visit the [main Mimir repository](https://github.com/ZebinX/Mimir-Uncertainty-Driving).
|
| 36 |
|
| 37 |
---
|
|
|
|
| 51 |
| `frame_mapping_navhard.yaml` | Frame index mapping for Navhard |
|
| 52 |
| `global_pose_navhard.npy` | Global pose data for Navhard |
|
| 53 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 54 |
## 📊 Results
|
| 55 |
|
| 56 |
+
<div align="center">
|
| 57 |
+
<img src="./assets/navhard.png" width="49%" />
|
| 58 |
+
<img src="./assets/navtest.png" width="49%" />
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 59 |
|
| 60 |
+
</div>
|
|
|
|
|
|
|
|
|
|
|
|
|
| 61 |
|
| 62 |
+
Please refer to the [paper](https://arxiv.org/pdf/2512.07130) for more details.
|
| 63 |
|
| 64 |
---
|
| 65 |
|
|
|
|
| 82 |
|
| 83 |
## Contact
|
| 84 |
|
| 85 |
+
If you have any questions or suggestions, please open an issue or contact: xzebin@bupt.edu.cn
|