dgrachev commited on
Commit
ebe3522
·
verified ·
1 Parent(s): 65dfb2b

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +88 -0
README.md ADDED
@@ -0,0 +1,88 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ tags:
4
+ - robotics
5
+ - manipulation
6
+ - simulation
7
+ - pybullet
8
+ ---
9
+
10
+ # A2 Environment Assets
11
+
12
+ Assets for the A2 tabletop manipulation environment (PyBullet simulation).
13
+
14
+ ## Contents
15
+
16
+ ### 3D Objects
17
+
18
+ - `simplified_objects/`: 66 simplified YCB objects for training
19
+ - `unseen_objects/`: 17 unseen objects for testing
20
+
21
+ ### Robot
22
+
23
+ - `ur5e/`: UR5e robot URDF and meshes
24
+ - `workspace/`: Workspace table and fixtures
25
+
26
+ ### Testing Cases
27
+
28
+ Pre-defined test scenarios for benchmark evaluation:
29
+
30
+ - `testing_cases/grasp_testing_cases/`: Grasp task test cases
31
+ - `seen/`: 10 cases with training objects
32
+ - `unseen/`: 5 cases with test objects
33
+ - `testing_cases/place_testing_cases/`: Place task test cases
34
+ - `seen/`: 20 cases
35
+ - `unseen/`: 10 cases
36
+ - `testing_cases/pp_testing_cases/`: Pick-and-place test cases
37
+ - `seen/`: 8 cases
38
+ - `unseen/`: 4 cases
39
+
40
+ ## Usage
41
+
42
+ Assets are automatically downloaded when using the A2 environment:
43
+
44
+ ```python
45
+ from lerobot.envs.a2 import A2Env
46
+
47
+ # Assets are downloaded to ~/.cache/a2_assets/
48
+ env = A2Env(task="grasp", num_objects=8)
49
+ ```
50
+
51
+ ### Manual Download
52
+
53
+ ```python
54
+ from huggingface_hub import snapshot_download
55
+
56
+ # Download all assets
57
+ snapshot_download(
58
+ repo_id="dgrachev/a2_assets",
59
+ repo_type="dataset",
60
+ local_dir="~/.cache/a2_assets"
61
+ )
62
+ ```
63
+
64
+ ## Test Case Format
65
+
66
+ Each test case file contains:
67
+ - Line 1: Language instruction (e.g., "grasp a round object")
68
+ - Line 2: Target object indices (space-separated)
69
+ - Lines 3+: Object URDF path and 6D pose (x, y, z, rx, ry, rz)
70
+
71
+ Example (`case00-round.txt`):
72
+ ```
73
+ grasp a round object
74
+ 0 1 2
75
+ assets/simplified_objects/005.urdf 0.61 0.04 0.2 4.22 0.45 1.39
76
+ assets/simplified_objects/012.urdf 0.53 -0.11 0.2 6.15 3.49 5.06
77
+ ...
78
+ ```
79
+
80
+ ## Related Resources
81
+
82
+ - [a2_pretrained](https://huggingface.co/dgrachev/a2_pretrained) - Pretrained A2 policy
83
+ - [lerobot_policy_a2](https://github.com/dgrachev/lerobot_policy_a2) - Policy package
84
+ - [lerobot_grach0v](https://github.com/grach0v/lerobot) - LeRobot fork with A2 environment
85
+
86
+ ## License
87
+
88
+ Apache-2.0