Instructions to use cdeplanne/microduck-test with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Microduck
How to use cdeplanne/microduck-test with Microduck:
# Replace SLOT with the slot specified in the model card (walk, stand, sitstand, ground_pick, kick_left, kick_right, roulade). sudo robotctl policy load SLOT cdeplanne/microduck-test
- Notebooks
- Google Colab
- Kaggle
Publish test from the Microduck Arena
Browse files- README.md +15 -0
- manifest.json +18 -0
- policy.onnx +3 -0
README.md
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: onnx
|
| 3 |
+
tags:
|
| 4 |
+
- microduck-policy
|
| 5 |
+
- microduck
|
| 6 |
+
- onnx
|
| 7 |
+
- robotics
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# microduck-test
|
| 11 |
+
|
| 12 |
+
`policy.onnx` is a Microduck policy: 61-value observation in, 14 joint
|
| 13 |
+
targets out, 50 Hz. Published from the Microduck Arena.
|
| 14 |
+
|
| 15 |
+
See `manifest.json` for the full contract.
|
manifest.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"schema_version": 2,
|
| 3 |
+
"model_api": 1,
|
| 4 |
+
"name": "test",
|
| 5 |
+
"kind": "perpetual",
|
| 6 |
+
"obs_len": 61,
|
| 7 |
+
"action_len": 14,
|
| 8 |
+
"action_scale": 1.0,
|
| 9 |
+
"entry_pose": "standing",
|
| 10 |
+
"robot": {
|
| 11 |
+
"model": "microduck",
|
| 12 |
+
"control_hz": 50
|
| 13 |
+
},
|
| 14 |
+
"published_by": "microduck-arena",
|
| 15 |
+
"arena": {
|
| 16 |
+
"event": "sprint-2m"
|
| 17 |
+
}
|
| 18 |
+
}
|
policy.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7a0d132f121d4bea3b713d3d7509500319389e0ac8de1ec9b390256471bbfc18
|
| 3 |
+
size 793771
|