mxlin043 commited on
Commit
1f6ce5e
·
verified ·
1 Parent(s): d3a44a2

Add files using upload-large-folder tool

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. README.md +156 -0
  2. SHA256SUMS +0 -0
  3. dataset/task/Brass_Streets/task_x-001342_y003501_z000098_t0_a.json +12 -0
  4. dataset/task/Brass_Streets/task_x-001342_y003501_z000098_t0_b.json +12 -0
  5. dataset/task/Brass_Streets/task_x-001342_y003501_z000098_t3_b.json +12 -0
  6. dataset/task/Brass_Streets/task_x-001593_y-000083_z000106_t3_c.json +12 -0
  7. dataset/task/Brass_Streets/task_x-001593_y-000083_z000106_t5_a_pd.json +23 -0
  8. dataset/task/Brass_Streets/task_x-001593_y-000083_z000106_t5_b_pd.json +23 -0
  9. dataset/task/Brass_Streets/task_x-001628_y003710_z000098_t1_a.json +12 -0
  10. dataset/task/Brass_Streets/task_x-001670_y000308_z000106_t0_a.json +12 -0
  11. dataset/task/Brass_Streets/task_x-001670_y000308_z000106_t0_b.json +12 -0
  12. dataset/task/Brass_Streets/task_x-001670_y000308_z000106_t5_a_p.json +23 -0
  13. dataset/task/Brass_Streets/task_x-001840_y-000471_z000108_t1_a.json +12 -0
  14. dataset/task/Brass_Streets/task_x-001855_y000030_z000106_t0_a.json +12 -0
  15. dataset/task/Brass_Streets/task_x-002073_y003542_z000098_t0_a.json +12 -0
  16. dataset/task/Brass_Streets/task_x-002073_y003542_z000098_t5_a_pd.json +23 -0
  17. dataset/task/Brass_Streets/task_x-002135_y001260_z000299_t0_b.json +12 -0
  18. dataset/task/Brass_Streets/task_x-002138_y-001220_z000099_t1_a.json +12 -0
  19. dataset/task/Brass_Streets/task_x-002944_y004394_z000098_t0_b.json +12 -0
  20. dataset/task/Brass_Streets/task_x-002944_y004394_z000098_t5_a_pd.json +23 -0
  21. dataset/task/Brass_Streets/task_x-004256_y003728_z000098_t3_a.json +12 -0
  22. dataset/task/Brass_Streets/task_x-004334_y003370_z000098_t2_a.json +12 -0
  23. dataset/task/Brass_Streets/task_x-004424_y002541_z000299_t5_a_p.json +23 -0
  24. dataset/task/Brass_Streets/task_x000024_y000201_z000099_t0_a.json +12 -0
  25. dataset/task/Brass_Streets/task_x000024_y000201_z000099_t5_a_p.json +23 -0
  26. dataset/task/Brass_Streets/task_x000024_y000201_z000099_t5_b_p.json +23 -0
  27. dataset/task/Brass_Streets/task_x002029_y003580_z-000302_t0_a.json +12 -0
  28. dataset/task/Brass_Streets/task_x002029_y003580_z-000302_t0_c.json +12 -0
  29. dataset/task/Brass_Streets/task_x002077_y003301_z-000302_t1_b.json +13 -0
  30. dataset/task/Brass_Streets/task_x002151_y003532_z-000302_t3_b.json +13 -0
  31. dataset/task/Brass_Streets/task_x002151_y003532_z-000302_t3_c.json +13 -0
  32. dataset/task/Brass_Streets/task_x002151_y003532_z-000302_t3_d.json +13 -0
  33. dataset/task/Brass_Streets/task_x002151_y003532_z-000302_t5_a_pd.json +24 -0
  34. dataset/task/Brass_Streets/task_x002262_y003542_z-000302_t1_b.json +12 -0
  35. dataset/task/Brass_Streets/task_x002822_y001598_z-000302_t3_a.json +12 -0
  36. dataset/task/Brass_Streets/task_x002822_y001598_z-000302_t3_b.json +12 -0
  37. dataset/task/Brass_Streets/task_x002822_y001598_z-000302_t3_c.json +12 -0
  38. dataset/task/Brass_Streets/task_x002822_y001598_z-000302_t3_d.json +12 -0
  39. dataset/task/Brass_Streets/task_x002822_y001598_z-000302_t3_e.json +12 -0
  40. dataset/task/Demonstration_Castle/task_x-044440_y016169_z-022522_t1_a.json +12 -0
  41. dataset/task/Demonstration_Castle/task_x-044440_y016169_z-022522_t1_b.json +12 -0
  42. dataset/task/Demonstration_Castle/task_x-044554_y015548_z-022529_t0_a.json +13 -0
  43. dataset/task/Demonstration_Castle/task_x-044554_y015548_z-022529_t0_b.json +13 -0
  44. dataset/task/Demonstration_Castle/task_x-044816_y016247_z-022529_t0_a.json +12 -0
  45. dataset/task/Demonstration_Castle/task_x-044816_y016247_z-022529_t0_b.json +12 -0
  46. dataset/task/Demonstration_Castle/task_x-044816_y016247_z-022529_t5_a_p.json +23 -0
  47. dataset/task/Demonstration_Castle/task_x-045917_y015877_z-022529_t2_a.json +12 -0
  48. dataset/task/Demonstration_Castle/task_x-045917_y015877_z-022529_t2_b.json +12 -0
  49. dataset/task/Demonstration_Castle/task_x-045917_y015877_z-022529_t2_c.json +12 -0
  50. dataset/task/Demonstration_Castle/task_x-046482_y015676_z-022523_t2_a.json +12 -0
README.md ADDED
@@ -0,0 +1,156 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: cc-by-nc-4.0
3
+ pretty_name: EmbRACE
4
+ language:
5
+ - en
6
+ tags:
7
+ - embodied-ai
8
+ - vision-language-navigation
9
+ - egocentric
10
+ - unreal-engine
11
+ - simulation
12
+ - benchmark
13
+ size_categories:
14
+ - 1K<n<10K
15
+ configs:
16
+ - config_name: default
17
+ data_files:
18
+ - split: dataset
19
+ path: data/dataset.parquet
20
+ - split: benchmark
21
+ path: data/benchmark.parquet
22
+ ---
23
+
24
+ # EmbRACE: Embodied Reasoning and Action in Complex Environments
25
+
26
+ [Viewer](https://huggingface.co/spaces/mxlin043/EmbRACE-Viewer) · [Paper](https://arxiv.org/abs/2507.10548)
27
+
28
+ EmbRACE consists of two parts. The dataset is a set of 3,421 human demonstrations for closed-loop tasks in 55 environments, 48,264 steps in all, in which every step is paired with a rationale written from the agent's viewpoint and verified by annotators. The benchmark is a set of 686 tasks in 7 further environments, run in the closed loop and scored on the agent's final position and on the door and object states.
29
+
30
+ | Part | Environments | Tasks | Steps | Rationales |
31
+ |---|---:|---:|---:|---:|
32
+ | dataset | 55 | 3,421 | 48,264 | 48,264 |
33
+ | benchmark | 7 | 686 | 10,552 | 0 |
34
+
35
+ | Type | Task type | dataset | benchmark |
36
+ |---:|---|---:|---:|
37
+ | 0 | Basic | 646 | 116 |
38
+ | 1 | Exploration | 591 | 117 |
39
+ | 2 | Dynamic Spatial-Semantic | 686 | 114 |
40
+ | 3 | Multi-stage | 644 | 106 |
41
+ | 4 | Open Door | 226 | 109 |
42
+ | 5 | Pick & Drop | 628 | 124 |
43
+
44
+ - **Basic**: The target is visible from the start pose and directly reachable.
45
+ - **Exploration**: The target is out of view at the start, so the agent searches the environment and commits to a direction before the target comes into view.
46
+ - **Dynamic Spatial-Semantic**: The target is visible but specified relationally or ordinally, for example "the second chair from the left". The relation is defined in the initial view.
47
+ - **Multi-stage**: The task names two targets to be reached in order.
48
+ - **Open Door**: The agent opens a door. Success is judged on the door state.
49
+ - **Pick & Drop**: The agent picks up an object and releases it beside a landmark. Success is judged on the object position.
50
+
51
+ ## Files
52
+
53
+ ```
54
+ dataset/task/<Env>/<Task>.json task files
55
+ dataset/trajectory/<Env>/<Task>/ frames and record of each demonstration
56
+ benchmark/task/<Env>/<Task>.json task files read by the evaluation code
57
+ benchmark/trajectory/<Env>/<Task>/ reference human demonstration of each task
58
+ benchmark/success_region/<Env>/<Task>/ success regions of each task
59
+ packed/{dataset,benchmark}/<Env>.tar all files of one environment in one archive
60
+ data/{dataset,benchmark}.parquet one row per task with its initial frame
61
+ index/web/tasks.json task index read by the viewer
62
+ environments.json the 62 environments
63
+ pd_objects.json the ten Pick & Drop objects
64
+ SHA256SUMS checksums of all files
65
+ ```
66
+
67
+ This is the layout that the archives in `packed/` hold and that the evaluation code reads. The files shown on this page are a copy for browsing, in which every task folder and task file carries the prefix `task_`, for example `benchmark/trajectory/CourtYard/task_x001312_y007194_z000118_t0_a/`, because the Hub does not accept some of the original names. `SHA256SUMS` lists every file under its name in the archives.
68
+
69
+ ## Download
70
+
71
+ The 62 archives hold every file under `dataset/` and `benchmark/`. Extracting them in one folder gives the layout above.
72
+
73
+ ```bash
74
+ hf download mxlin043/EmbRACE --repo-type dataset --local-dir EmbRACE \
75
+ --include "packed/*" --include "environments.json" --include "pd_objects.json"
76
+ cd EmbRACE && for f in packed/*/*.tar; do tar -xf "$f"; done
77
+ ```
78
+
79
+ One environment can also be downloaded on its own, for example
80
+
81
+ ```bash
82
+ hf download mxlin043/EmbRACE --repo-type dataset --local-dir EmbRACE --include "packed/benchmark/CourtYard.tar"
83
+ ```
84
+
85
+ The task table loads with the `datasets` library.
86
+
87
+ ```python
88
+ from datasets import load_dataset
89
+
90
+ tasks = load_dataset("mxlin043/EmbRACE", split="dataset")
91
+ ```
92
+
93
+ Running the benchmark requires the UnrealZoo UE5.6 package of version v3.0.2 and the evaluation code.
94
+
95
+ ## Format
96
+
97
+ ### Task files
98
+
99
+ | Field | Content |
100
+ |---|---|
101
+ | `Instruction` | the instruction |
102
+ | `Type` | the task type, 0 to 5 |
103
+ | `Start_Pose` | position and orientation of the agent at the start, `[x, y, z, pitch, yaw, roll]` |
104
+ | `PD_ID`, `PD_Size` | Pick & Drop only, the object and its scale |
105
+ | `PD_Type` | Pick & Drop only, `pd` when the object is released beside a landmark and `p` when the task ends with the object held |
106
+ | `PD_Start_Pose` | Pick & Drop only, the pose at which the object is placed |
107
+ | `Exposure` | optional, an exposure offset for rendering that replaces the default of the environment |
108
+
109
+ A task is named by its start position in centimeters, its type and a letter, for example `x001312_y007194_z000118_t0_a`. The names of Pick & Drop tasks end in `_p` or `_pd` following `PD_Type`.
110
+
111
+ ### Trajectory records
112
+
113
+ Each demonstration is one folder with its frames `000.jpg`, `001.jpg`, … at 640×480 and the record `infos.json`. The frames are the egocentric frame before every action and the frame after the last one, so a demonstration of n actions has n+1 frames. The pose and state streams have n+1 entries aligned with the frames. Poses are given in the coordinate frame of the environment, in centimeters and degrees.
114
+
115
+ | Field | Content |
116
+ |---|---|
117
+ | `Instruction`, `Type`, `Start_Pose`, `PD_ID`, `PD_Size`, `PD_Type`, `PD_Start_Pose` | as in the task file |
118
+ | `Action` | the demonstrated actions (n) |
119
+ | `Pose_Trajectory` | the agent pose (n+1) |
120
+ | `Door_State_Trajectory` | Open Door only, the door state, 0 closed and 1 open (n+1) |
121
+ | `PD_Pose_Trajectory` | Pick & Drop only, the object pose (n+1) |
122
+ | `PD_Dist_2D` | Pick & Drop only, the horizontal distance from the agent to the object (n+1) |
123
+ | `Rationale` | dataset only, the rationale of each step (n) |
124
+
125
+ The actions are `MoveForward`, `MoveBackward`, `TurnLeft`, `TurnRight`, `LookUp`, `LookDown`, `OpenDoor`, `Pick`, `Drop`, `MidwayTarget` and `Finish`. `MidwayTarget` declares arrival at the first target of a Multi-stage task.
126
+
127
+ ### Success regions
128
+
129
+ Each benchmark task has an `area.json` with the regions it is scored against. Two images show the regions, `xy_region.png` as a top-down plot and `ue_top_down_overlay.jpg` drawn over the scene. Multi-stage tasks have one pair per target, prefixed `midway_` and `final_`.
130
+
131
+ | Field | Content |
132
+ |---|---|
133
+ | `coordinate_system` | the frame of the geometry, the XY plane of the environment in centimeters |
134
+ | `distance_metric` | distances are measured in the XY plane |
135
+ | `stages` | one entry per target, `final` and for Multi-stage tasks also `midway` |
136
+ | `character_regions` | the polygons in which the agent has to stand |
137
+ | `target_footprints` | the footprint of the target |
138
+ | `character_reference_z` | the reference height of the agent at the target |
139
+ | `pd_object_regions`, `pd_object_reference_z` | Pick & Drop only, the region and the reference height of the object |
140
+ | `shortest_path_uu` | the reference length of SPL, in centimeters |
141
+
142
+ ### Other files
143
+
144
+ `environments.json` gives each environment its identifier, its name in the paper, its part and its setting (`id`, `name`, `split`, `setting`). `pd_objects.json` maps `PD_ID` to the name of the object. The Parquet tables in `data/` hold one row per task with its environment, type, instruction, initial frame, actions and rationales.
145
+
146
+ ## License
147
+
148
+ The data are released under the [Creative Commons Attribution-NonCommercial 4.0](LICENSE) license, for research use with attribution. The environments are those of UnrealZoo and are obtained from its release under its terms.
149
+
150
+ ## Citation
151
+
152
+ To be added.
153
+
154
+ ## Changelog
155
+
156
+ This is version 1.0. Corrections to the data are recorded in `CHANGELOG.md`.
SHA256SUMS ADDED
The diff for this file is too large to render. See raw diff
 
dataset/task/Brass_Streets/task_x-001342_y003501_z000098_t0_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Walk to the doorway on the patterned building.",
3
+ "Type": 0,
4
+ "Start_Pose": [
5
+ -1342.141,
6
+ 3500.513,
7
+ 98.25,
8
+ 0,
9
+ 144.96,
10
+ 0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x-001342_y003501_z000098_t0_b.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Approach the small stone shrine with the red gem emblem.",
3
+ "Type": 0,
4
+ "Start_Pose": [
5
+ -1342.141,
6
+ 3500.513,
7
+ 98.25,
8
+ 0.0,
9
+ 144.96,
10
+ -0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x-001342_y003501_z000098_t3_b.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Head toward the red-jeweled emblem on the stone structure, then move to the arched doorway entrance.",
3
+ "Type": 3,
4
+ "Start_Pose": [
5
+ -1342.141,
6
+ 3500.513,
7
+ 98.25,
8
+ 0.0,
9
+ 144.96,
10
+ -0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x-001593_y-000083_z000106_t3_c.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Approach the small square lattice window, then move to the round decorative emblem on the wall.",
3
+ "Type": 3,
4
+ "Start_Pose": [
5
+ -1593.315,
6
+ -82.524,
7
+ 106.284,
8
+ 0.0,
9
+ -63.36,
10
+ 0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x-001593_y-000083_z000106_t5_a_pd.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Grab the jerry can at the arched window with iron lattice, then place it near the red hanging banner.",
3
+ "Type": 5,
4
+ "Start_Pose": [
5
+ -1593.315,
6
+ -82.524,
7
+ 106.284,
8
+ 0,
9
+ -63.36,
10
+ 0
11
+ ],
12
+ "PD_ID": 13,
13
+ "PD_Size": 1.5,
14
+ "PD_Type": "pd",
15
+ "PD_Start_Pose": [
16
+ -1430.885,
17
+ -1507.982,
18
+ 40.662,
19
+ 0,
20
+ -30,
21
+ 0
22
+ ]
23
+ }
dataset/task/Brass_Streets/task_x-001593_y-000083_z000106_t5_b_pd.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Lift the wicker basket by the round decorative emblem on the wall, then leave it near the arched window with iron lattice.",
3
+ "Type": 5,
4
+ "Start_Pose": [
5
+ -1593.315,
6
+ -82.524,
7
+ 106.284,
8
+ 0,
9
+ -63.36,
10
+ 0
11
+ ],
12
+ "PD_ID": 7,
13
+ "PD_Size": 2.5,
14
+ "PD_Type": "pd",
15
+ "PD_Start_Pose": [
16
+ -881.858,
17
+ -930.298,
18
+ 48.65,
19
+ 90,
20
+ -43.36,
21
+ 0
22
+ ]
23
+ }
dataset/task/Brass_Streets/task_x-001628_y003710_z000098_t1_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Search for the open building entrance beneath the zigzag-patterned trim, then approach it.",
3
+ "Type": 1,
4
+ "Start_Pose": [
5
+ -1628.098,
6
+ 3710.196,
7
+ 98.25,
8
+ 0.0,
9
+ -131.52,
10
+ 0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x-001670_y000308_z000106_t0_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Walk to the hanging red banner on the building.",
3
+ "Type": 0,
4
+ "Start_Pose": [
5
+ -1669.565,
6
+ 307.789,
7
+ 106.284,
8
+ 0.0,
9
+ -77.76,
10
+ 0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x-001670_y000308_z000106_t0_b.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Head toward the round medallion emblem on the wall.",
3
+ "Type": 0,
4
+ "Start_Pose": [
5
+ -1669.565,
6
+ 307.789,
7
+ 106.284,
8
+ 0.0,
9
+ -77.76,
10
+ 0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x-001670_y000308_z000106_t5_a_p.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Lift the trash bag beside the round medallion emblem on the wall.",
3
+ "Type": 5,
4
+ "Start_Pose": [
5
+ -1669.565,
6
+ 307.789,
7
+ 106.284,
8
+ 0,
9
+ -77.76,
10
+ 0
11
+ ],
12
+ "PD_ID": 20,
13
+ "PD_Size": 1.5,
14
+ "PD_Type": "p",
15
+ "PD_Start_Pose": [
16
+ -912.701,
17
+ -958.87,
18
+ 28.697,
19
+ 0,
20
+ 0,
21
+ 0
22
+ ]
23
+ }
dataset/task/Brass_Streets/task_x-001840_y-000471_z000108_t1_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Search for the large arched window with an ornate lattice pattern that has three smaller lattice windows arranged around it on the stone wall; after locating it, head over to that window.",
3
+ "Type": 1,
4
+ "Start_Pose": [
5
+ -1840.045,
6
+ -470.87,
7
+ 107.575,
8
+ 0.0,
9
+ 156.48,
10
+ -0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x-001855_y000030_z000106_t0_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Walk to the arched window with lattice grillwork near the ground.",
3
+ "Type": 0,
4
+ "Start_Pose": [
5
+ -1855.332,
6
+ 29.534,
7
+ 106.282,
8
+ 0.0,
9
+ 23.04,
10
+ -0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x-002073_y003542_z000098_t0_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Walk to the dark doorway beneath the zigzag-patterned wall.",
3
+ "Type": 0,
4
+ "Start_Pose": [
5
+ -2072.602,
6
+ 3541.841,
7
+ 98.25,
8
+ 0.0,
9
+ 95.04,
10
+ -0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x-002073_y003542_z000098_t5_a_pd.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Pick up the candelabra at the dark wooden doorway, then set it down near the broken arched lattice window.",
3
+ "Type": 5,
4
+ "Start_Pose": [
5
+ -2072.602,
6
+ 3541.841,
7
+ 98.25,
8
+ 0,
9
+ 100.8,
10
+ 0
11
+ ],
12
+ "PD_ID": 17,
13
+ "PD_Size": 1.7,
14
+ "PD_Type": "pd",
15
+ "PD_Start_Pose": [
16
+ -1971.889,
17
+ 4165.949,
18
+ 20.25,
19
+ 0,
20
+ 0,
21
+ 0
22
+ ]
23
+ }
dataset/task/Brass_Streets/task_x-002135_y001260_z000299_t0_b.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Approach the small square window opening in the stone wall.",
3
+ "Type": 0,
4
+ "Start_Pose": [
5
+ -2134.543,
6
+ 1260.224,
7
+ 298.648,
8
+ 0.0,
9
+ -131.52,
10
+ 0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x-002138_y-001220_z000099_t1_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Locate the stone pedestal base of the large jewel-encrusted golden dome structure, then navigate to it.",
3
+ "Type": 1,
4
+ "Start_Pose": [
5
+ -2138.02,
6
+ -1220.425,
7
+ 98.529,
8
+ 0.0,
9
+ -112.32,
10
+ 0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x-002944_y004394_z000098_t0_b.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Approach the pair of small connected windows on the stone wall.",
3
+ "Type": 0,
4
+ "Start_Pose": [
5
+ -2943.532,
6
+ 4394.476,
7
+ 98.25,
8
+ 0.0,
9
+ -144.0,
10
+ 0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x-002944_y004394_z000098_t5_a_pd.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Lift the metal bucket at the base of the stairs, then leave it by the ground-level arched wooden window.",
3
+ "Type": 5,
4
+ "Start_Pose": [
5
+ -2943.532,
6
+ 4394.476,
7
+ 98.25,
8
+ 0,
9
+ -144,
10
+ 0
11
+ ],
12
+ "PD_ID": 18,
13
+ "PD_Size": 2.0,
14
+ "PD_Type": "pd",
15
+ "PD_Start_Pose": [
16
+ -3576.034,
17
+ 3866.429,
18
+ 23.25,
19
+ 0,
20
+ 0,
21
+ 0
22
+ ]
23
+ }
dataset/task/Brass_Streets/task_x-004256_y003728_z000098_t3_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Walk to the large arched window with red lattice, then approach the two small square vents.",
3
+ "Type": 3,
4
+ "Start_Pose": [
5
+ -4255.845,
6
+ 3727.756,
7
+ 98.25,
8
+ 0.0,
9
+ 157.44,
10
+ -0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x-004334_y003370_z000098_t2_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Walk to the leftmost ground-floor arched window.",
3
+ "Type": 2,
4
+ "Start_Pose": [
5
+ -4334.336,
6
+ 3370.254,
7
+ 98.25,
8
+ 0.0,
9
+ 68.16,
10
+ -0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x-004424_y002541_z000299_t5_a_p.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Grab the trash bag at the golden door beneath the ornate archway.",
3
+ "Type": 5,
4
+ "Start_Pose": [
5
+ -4424.384,
6
+ 2540.654,
7
+ 298.65,
8
+ 0,
9
+ 179.52,
10
+ 0
11
+ ],
12
+ "PD_ID": 20,
13
+ "PD_Size": 1.5,
14
+ "PD_Type": "p",
15
+ "PD_Start_Pose": [
16
+ -5763.464,
17
+ 2439.462,
18
+ 205.06,
19
+ -0.138,
20
+ -101.609,
21
+ -21.174
22
+ ]
23
+ }
dataset/task/Brass_Streets/task_x000024_y000201_z000099_t0_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Walk to the red column between the arched windows.",
3
+ "Type": 0,
4
+ "Start_Pose": [
5
+ 24.313,
6
+ 201.366,
7
+ 98.615,
8
+ 0.0,
9
+ -12.48,
10
+ 0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x000024_y000201_z000099_t5_a_p.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Pick up the tomato near the red column between the arched windows.",
3
+ "Type": 5,
4
+ "Start_Pose": [
5
+ 24.313,
6
+ 201.366,
7
+ 98.615,
8
+ 0,
9
+ -12.48,
10
+ 0
11
+ ],
12
+ "PD_ID": 0,
13
+ "PD_Size": 8.0,
14
+ "PD_Type": "p",
15
+ "PD_Start_Pose": [
16
+ 1000.52,
17
+ -14.696,
18
+ 68.65,
19
+ 0,
20
+ 0,
21
+ 0
22
+ ]
23
+ }
dataset/task/Brass_Streets/task_x000024_y000201_z000099_t5_b_p.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Grab the tomato at the base of the tall pale pillar.",
3
+ "Type": 5,
4
+ "Start_Pose": [
5
+ 24.313,
6
+ 201.366,
7
+ 98.615,
8
+ 0,
9
+ -12.48,
10
+ 0
11
+ ],
12
+ "PD_ID": 0,
13
+ "PD_Size": 8.0,
14
+ "PD_Type": "p",
15
+ "PD_Start_Pose": [
16
+ 1108.148,
17
+ 561.967,
18
+ 33.869,
19
+ -42.307,
20
+ 102.115,
21
+ 86.545
22
+ ]
23
+ }
dataset/task/Brass_Streets/task_x002029_y003580_z-000302_t0_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Walk to the ornate golden pedestal with the hanging lantern.",
3
+ "Type": 0,
4
+ "Start_Pose": [
5
+ 2029.388,
6
+ 3579.671,
7
+ -301.75,
8
+ 0.0,
9
+ -81.6,
10
+ 0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x002029_y003580_z-000302_t0_c.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Move to the base of the white stone column.",
3
+ "Type": 0,
4
+ "Start_Pose": [
5
+ 2029.388,
6
+ 3579.671,
7
+ -301.75,
8
+ 0.0,
9
+ -81.6,
10
+ 0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x002077_y003301_z-000302_t1_b.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Once you locate the bottom edge of the long red banner hanging against the stone wall, navigate to reach it.",
3
+ "Type": 1,
4
+ "Start_Pose": [
5
+ 2077.433,
6
+ 3301.052,
7
+ -301.75,
8
+ 0.0,
9
+ 77.76,
10
+ -0.0
11
+ ],
12
+ "Exposure": 0.5
13
+ }
dataset/task/Brass_Streets/task_x002151_y003532_z-000302_t3_b.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Head toward the red hanging banner, then move to the stone steps.",
3
+ "Type": 3,
4
+ "Start_Pose": [
5
+ 2150.882,
6
+ 3531.736,
7
+ -301.75,
8
+ 0.0,
9
+ -87.36,
10
+ 0.0
11
+ ],
12
+ "Exposure": 0.5
13
+ }
dataset/task/Brass_Streets/task_x002151_y003532_z-000302_t3_c.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Approach the ornate golden pillar base, then reach the white pillar.",
3
+ "Type": 3,
4
+ "Start_Pose": [
5
+ 2150.882,
6
+ 3531.736,
7
+ -301.75,
8
+ 0.0,
9
+ -87.36,
10
+ 0.0
11
+ ],
12
+ "Exposure": 0.5
13
+ }
dataset/task/Brass_Streets/task_x002151_y003532_z-000302_t3_d.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Navigate to the stone steps, then proceed to the red hanging banner.",
3
+ "Type": 3,
4
+ "Start_Pose": [
5
+ 2150.882,
6
+ 3531.736,
7
+ -301.75,
8
+ 0.0,
9
+ -87.36,
10
+ 0.0
11
+ ],
12
+ "Exposure": 0.5
13
+ }
dataset/task/Brass_Streets/task_x002151_y003532_z-000302_t5_a_pd.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Grab the ceramic pot at the ornate golden pillar base, then place it near the stone steps.",
3
+ "Type": 5,
4
+ "Start_Pose": [
5
+ 2150.882,
6
+ 3531.736,
7
+ -301.75,
8
+ 0,
9
+ -87.36,
10
+ 0
11
+ ],
12
+ "PD_ID": 8,
13
+ "PD_Size": 1.5,
14
+ "PD_Type": "pd",
15
+ "PD_Start_Pose": [
16
+ 2457.184,
17
+ 2791.049,
18
+ -376.75,
19
+ 0,
20
+ 0,
21
+ 0
22
+ ],
23
+ "Exposure": 0.5
24
+ }
dataset/task/Brass_Streets/task_x002262_y003542_z-000302_t1_b.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Once you locate the single arched window fitted with a carved wooden lattice screen near ground level, approach it.",
3
+ "Type": 1,
4
+ "Start_Pose": [
5
+ 2261.68,
6
+ 3541.733,
7
+ -301.75,
8
+ 0.0,
9
+ -24.0,
10
+ 0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x002822_y001598_z-000302_t3_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Walk to the ornate golden lattice pillar, then approach the white stone column.",
3
+ "Type": 3,
4
+ "Start_Pose": [
5
+ 2821.69,
6
+ 1598.149,
7
+ -301.75,
8
+ 0.0,
9
+ 91.2,
10
+ -0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x002822_y001598_z-000302_t3_b.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Head toward the base of the white column, then reach the staircase steps.",
3
+ "Type": 3,
4
+ "Start_Pose": [
5
+ 2821.69,
6
+ 1598.149,
7
+ -301.75,
8
+ 0.0,
9
+ 91.2,
10
+ -0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x002822_y001598_z-000302_t3_c.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Approach the carved golden pillar, then move to the staircase steps.",
3
+ "Type": 3,
4
+ "Start_Pose": [
5
+ 2821.69,
6
+ 1598.149,
7
+ -301.75,
8
+ 0.0,
9
+ 91.2,
10
+ -0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x002822_y001598_z-000302_t3_d.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Navigate to the stone staircase, then proceed to the golden lattice pillar.",
3
+ "Type": 3,
4
+ "Start_Pose": [
5
+ 2821.69,
6
+ 1598.149,
7
+ -301.75,
8
+ 0.0,
9
+ 91.2,
10
+ -0.0
11
+ ]
12
+ }
dataset/task/Brass_Streets/task_x002822_y001598_z-000302_t3_e.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Move to the white column, then walk to the ornate golden pillar.",
3
+ "Type": 3,
4
+ "Start_Pose": [
5
+ 2821.69,
6
+ 1598.149,
7
+ -301.75,
8
+ 0.0,
9
+ 91.2,
10
+ -0.0
11
+ ]
12
+ }
dataset/task/Demonstration_Castle/task_x-044440_y016169_z-022522_t1_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Search for the base of the stone staircase where it meets the sandy ground floor; after locating it, navigate to that spot.",
3
+ "Type": 1,
4
+ "Start_Pose": [
5
+ -44439.748,
6
+ 16169.339,
7
+ -22522.077,
8
+ 0.0,
9
+ -59.84,
10
+ 0.0
11
+ ]
12
+ }
dataset/task/Demonstration_Castle/task_x-044440_y016169_z-022522_t1_b.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Locate the ground-level doorway that is half-covered by a metal grid gate and half open; once you have identified it, move toward that entrance.",
3
+ "Type": 1,
4
+ "Start_Pose": [
5
+ -44439.748,
6
+ 16169.339,
7
+ -22522.077,
8
+ 0.0,
9
+ -59.84,
10
+ 0.0
11
+ ]
12
+ }
dataset/task/Demonstration_Castle/task_x-044554_y015548_z-022529_t0_a.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Walk to the pointed stone pillar in the sandy courtyard.",
3
+ "Type": 0,
4
+ "Start_Pose": [
5
+ -44553.703,
6
+ 15548.054,
7
+ -22529.288,
8
+ 0.0,
9
+ -151.68,
10
+ 0.0
11
+ ],
12
+ "Exposure": -1.5
13
+ }
dataset/task/Demonstration_Castle/task_x-044554_y015548_z-022529_t0_b.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Approach the white cloth-draped object near the ruins.",
3
+ "Type": 0,
4
+ "Start_Pose": [
5
+ -44553.703,
6
+ 15548.054,
7
+ -22529.288,
8
+ 0.0,
9
+ -151.68,
10
+ 0.0
11
+ ],
12
+ "Exposure": -1.5
13
+ }
dataset/task/Demonstration_Castle/task_x-044816_y016247_z-022529_t0_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Walk to the entrance of the pointed stone archway.",
3
+ "Type": 0,
4
+ "Start_Pose": [
5
+ -44815.535,
6
+ 16246.692,
7
+ -22529.287,
8
+ 0.0,
9
+ -0.0,
10
+ 0.0
11
+ ]
12
+ }
dataset/task/Demonstration_Castle/task_x-044816_y016247_z-022529_t0_b.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Head toward the bottom of the wooden staircase.",
3
+ "Type": 0,
4
+ "Start_Pose": [
5
+ -44815.535,
6
+ 16246.692,
7
+ -22529.287,
8
+ 0.0,
9
+ -0.0,
10
+ 0.0
11
+ ]
12
+ }
dataset/task/Demonstration_Castle/task_x-044816_y016247_z-022529_t5_a_p.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Grab the wicker basket at the bottom of the wooden staircase.",
3
+ "Type": 5,
4
+ "Start_Pose": [
5
+ -44815.535,
6
+ 16246.692,
7
+ -22529.287,
8
+ 0,
9
+ 0,
10
+ 0
11
+ ],
12
+ "PD_ID": 7,
13
+ "PD_Size": 2.5,
14
+ "PD_Type": "p",
15
+ "PD_Start_Pose": [
16
+ -43876.738,
17
+ 16695.199,
18
+ -22579.287,
19
+ 90,
20
+ 19.999,
21
+ 0
22
+ ]
23
+ }
dataset/task/Demonstration_Castle/task_x-045917_y015877_z-022529_t2_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Walk to the small pedestal on the far left.",
3
+ "Type": 2,
4
+ "Start_Pose": [
5
+ -45917.285,
6
+ 15876.847,
7
+ -22529.288,
8
+ 0.0,
9
+ 140.16,
10
+ -0.0
11
+ ]
12
+ }
dataset/task/Demonstration_Castle/task_x-045917_y015877_z-022529_t2_b.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Head toward the cross-shaped headstone right of center.",
3
+ "Type": 2,
4
+ "Start_Pose": [
5
+ -45917.285,
6
+ 15876.847,
7
+ -22529.288,
8
+ 0.0,
9
+ 140.16,
10
+ -0.0
11
+ ]
12
+ }
dataset/task/Demonstration_Castle/task_x-045917_y015877_z-022529_t2_c.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Move to the leftmost arched window with lattice grille.",
3
+ "Type": 2,
4
+ "Start_Pose": [
5
+ -45917.285,
6
+ 15876.847,
7
+ -22529.288,
8
+ 0.0,
9
+ 140.16,
10
+ -0.0
11
+ ]
12
+ }
dataset/task/Demonstration_Castle/task_x-046482_y015676_z-022523_t2_a.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Instruction": "Approach the leftmost arched window with the grille.",
3
+ "Type": 2,
4
+ "Start_Pose": [
5
+ -46481.583,
6
+ 15676.37,
7
+ -22522.651,
8
+ 0.0,
9
+ 178.56,
10
+ -0.0
11
+ ]
12
+ }