Hugging Face's logo

Datasets:
TESS-Computer
/
carla-simlingo-raw

Tasks:
Robotics
Image-to-Text
Modalities:
Tabular
Text
Formats:
parquet
Size:
100K - 1M
Tags:
autonomous-driving
carla
simlingo
behavioral-cloning
Libraries:
Datasets
Dask
Polars
License:
Dataset card Data Studio Files
xet
Community
1
HusseinLezzaik commited on
Commit
b5e0d04
·
verified ·
1 Parent(s): 56fe06d

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +139 -0
README.md ADDED
@@ -0,0 +1,139 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ task_categories:
4
+ - robotics
5
+ - image-to-text
6
+ tags:
7
+ - autonomous-driving
8
+ - carla
9
+ - simlingo
10
+ - behavioral-cloning
11
+ size_categories:
12
+ - 100K<n<1M
13
+ ---
14
+
15
+ # SimLingo CARLA Dataset (Raw, 4Hz)
16
+
17
+ Raw driving data from CARLA simulator. No transformations or derived fields - all original measurements preserved as-is.
18
+
19
+ ## Dataset Summary
20
+
21
+ - **Source**: [SimLingo](https://huggingface.co/datasets/RenzKa/simlingo) (CVPR 2025)
22
+ - **Scale**: 228,757 frames (23 shards)
23
+ - **Frame Rate**: 4 FPS
24
+ - **Resolution**: 1024x512 RGB
25
+ - **Routes**: Complete driving episodes (routes never split across shards)
26
+
27
+ ## Column Schema
28
+
29
+ ### Core Fields
30
+ | Column | Type | Description |
31
+ |--------|------|-------------|
32
+ | `route_id` | string | Route identifier |
33
+ | `frame_idx` | int32 | Frame index within route |
34
+ | `image` | bytes | Original JPEG image bytes |
35
+
36
+ ### Control Signals (Raw)
37
+ | Column | Type | Description |
38
+ |--------|------|-------------|
39
+ | `steer` | float32 | Steering [-1, 1] |
40
+ | `throttle` | float32 | Throttle [0, 1] |
41
+ | `brake` | bool | Brake applied |
42
+
43
+ ### Vehicle State
44
+ | Column | Type | Description |
45
+ |--------|------|-------------|
46
+ | `speed` | float32 | Current speed (m/s) |
47
+ | `target_speed` | float32 | Target speed |
48
+ | `speed_limit` | float32 | Speed limit |
49
+ | `theta` | float32 | Heading angle |
50
+ | `angle` | float32 | Angle to target |
51
+
52
+ ### Navigation
53
+ | Column | Type | Description |
54
+ |--------|------|-------------|
55
+ | `command` | int32 | Navigation command |
56
+ | `next_command` | int32 | Next navigation command |
57
+ | `pos_global` | string (JSON) | Global position [x, y] |
58
+ | `target_point` | string (JSON) | Target point |
59
+ | `target_point_next` | string (JSON) | Next target point |
60
+ | `aim_wp` | string (JSON) | Aim waypoint |
61
+ | `route` | string (JSON) | Planned route waypoints |
62
+ | `route_original` | string (JSON) | Original route waypoints |
63
+ | `changed_route` | bool | Route was changed |
64
+
65
+ ### Hazards & Environment
66
+ | Column | Type | Description |
67
+ |--------|------|-------------|
68
+ | `junction` | bool | In junction |
69
+ | `vehicle_hazard` | bool | Vehicle hazard detected |
70
+ | `vehicle_affecting_id` | int32 | ID of affecting vehicle |
71
+ | `walker_hazard` | bool | Pedestrian hazard |
72
+ | `walker_affecting_id` | int32 | ID of affecting pedestrian |
73
+ | `light_hazard` | bool | Traffic light hazard |
74
+ | `stop_sign_hazard` | bool | Stop sign hazard |
75
+ | `stop_sign_close` | bool | Stop sign nearby |
76
+ | `walker_close` | bool | Pedestrian nearby |
77
+ | `walker_close_id` | int32 | ID of nearby pedestrian |
78
+ | `speed_reduced_by_obj_type` | string | Object type causing speed reduction |
79
+ | `speed_reduced_by_obj_id` | int32 | Object ID causing speed reduction |
80
+ | `speed_reduced_by_obj_distance` | float32 | Distance to speed-reducing object |
81
+ | `control_brake` | bool | Control brake applied |
82
+
83
+ ### Augmentation (from SimLingo)
84
+ | Column | Type | Description |
85
+ |--------|------|-------------|
86
+ | `augmentation_translation` | float32 | Translation augmentation |
87
+ | `augmentation_rotation` | float32 | Rotation augmentation |
88
+
89
+ ### Transforms
90
+ | Column | Type | Description |
91
+ |--------|------|-------------|
92
+ | `ego_matrix` | string (JSON) | 4x4 ego vehicle transform matrix |
93
+ | `boxes` | string (JSON) | 3D bounding boxes for all objects |
94
+
95
+ ### Commentary (Optional)
96
+ | Column | Type | Description |
97
+ |--------|------|-------------|
98
+ | `commentary` | string | Natural language commentary |
99
+ | `commentary_data` | string (JSON) | Full commentary object with metadata |
100
+
101
+ ## Usage
102
+
103
+ ```python
104
+ from datasets import load_dataset
105
+ import json
106
+
107
+ ds = load_dataset("TESS-Computer/carla-simlingo-raw", split="train")
108
+
109
+ sample = ds[0]
110
+ print(sample['route_id'])
111
+ print(sample['steer'], sample['throttle'], sample['brake'])
112
+ print(sample['speed'])
113
+
114
+ # Parse JSON fields
115
+ pos = json.loads(sample['pos_global'])
116
+ boxes = json.loads(sample['boxes']) if sample['boxes'] else []
117
+ ```
118
+
119
+ ## Data Collection
120
+
121
+ - **Simulator**: CARLA 0.9.15 (Leaderboard 2.0)
122
+ - **Expert**: PDM-Lite (rule-based, 100% route completion)
123
+ - **Scenarios**: Single-scenario routes with random weather
124
+ - **Towns**: Towns 1-13
125
+
126
+ ## Citation
127
+
128
+ ```bibtex
129
+ @inproceedings{renz2025simlingo,
130
+ title={SimLingo: Vision-Only Closed-Loop Autonomous Driving with Language-Action Alignment},
131
+ author={Renz, Katrin and Chen, Long and Arani, Elahe and Sinavski, Oleg},
132
+ booktitle={CVPR},
133
+ year={2025},
134
+ }
135
+ ```
136
+
137
+ ## License
138
+
139
+ MIT (dataset processing code). Original data subject to [SimLingo](https://huggingface.co/datasets/RenzKa/simlingo) and [CARLA](https://carla.org/) licenses.