Hugging Face's logo

Datasets:
YongshuoLiu
/
LANTERN

Tasks:
Other
Languages:
English
Size:
100K<n<1M
Tags:
autonomous-driving
cooperative-driving
vision-language-action
carla
closed-loop
License:
Dataset card Data Studio Files
xet
 Community
Dataset Preview
Duplicate
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
The dataset generation failed
Error code:   DatasetGenerationError
Exception:    CastError
Message:      Couldn't cast
valid: bool
route_a_count: int64
route_b_count: int64
paired_count: int64
family_counts: struct<occluded_pedestrian: int64, right_turn_on_red_yield: int64, static_obstacle_reveal: int64, ve (... 77 chars omitted)
  child 0, occluded_pedestrian: int64
  child 1, right_turn_on_red_yield: int64
  child 2, static_obstacle_reveal: int64
  child 3, vehicle_cut_in: int64
  child 4, wrong_way_left_turn: int64
  child 5, wrong_way_right_turn: int64
town_counts: struct<Town01: int64, Town02: int64, Town03: int64, Town04: int64, Town05: int64, Town06: int64, Tow (... 73 chars omitted)
  child 0, Town01: int64
  child 1, Town02: int64
  child 2, Town03: int64
  child 3, Town04: int64
  child 4, Town05: int64
  child 5, Town06: int64
  child 6, Town07: int64
  child 7, Town10HD: int64
  child 8, Town12: int64
  child 9, Town13: int64
  child 10, Town15: int64
hazard_behaviors_in_route_b: int64
context_preserving_control_scenarios: int64
cooperative_hints_in_route_b: int64
route_geometry_weather_mismatches: int64
manifest: string
lock: string
frame: int64
exclusion_reason: string
supervision_end_frame: int64
warning_on_frame: int64
canonical_id: string
prediction_horizon_frames: int64
excluded: bool
to
{'canonical_id': Value('string'), 'excluded': Value('bool'), 'exclusion_reason': Value('string'), 'frame': Value('int64'), 'prediction_horizon_frames': Value('int64'), 'supervision_end_frame': Value('int64'), 'warning_on_frame': Value('int64')}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1827, in _prepare_split_single
                  for key, table in generator:
                                    ^^^^^^^^^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 613, in wrapped
                  for item in generator(*args, **kwargs):
                              ~~~~~~~~~^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              valid: bool
              route_a_count: int64
              route_b_count: int64
              paired_count: int64
              family_counts: struct<occluded_pedestrian: int64, right_turn_on_red_yield: int64, static_obstacle_reveal: int64, ve (... 77 chars omitted)
                child 0, occluded_pedestrian: int64
                child 1, right_turn_on_red_yield: int64
                child 2, static_obstacle_reveal: int64
                child 3, vehicle_cut_in: int64
                child 4, wrong_way_left_turn: int64
                child 5, wrong_way_right_turn: int64
              town_counts: struct<Town01: int64, Town02: int64, Town03: int64, Town04: int64, Town05: int64, Town06: int64, Tow (... 73 chars omitted)
                child 0, Town01: int64
                child 1, Town02: int64
                child 2, Town03: int64
                child 3, Town04: int64
                child 4, Town05: int64
                child 5, Town06: int64
                child 6, Town07: int64
                child 7, Town10HD: int64
                child 8, Town12: int64
                child 9, Town13: int64
                child 10, Town15: int64
              hazard_behaviors_in_route_b: int64
              context_preserving_control_scenarios: int64
              cooperative_hints_in_route_b: int64
              route_geometry_weather_mismatches: int64
              manifest: string
              lock: string
              frame: int64
              exclusion_reason: string
              supervision_end_frame: int64
              warning_on_frame: int64
              canonical_id: string
              prediction_horizon_frames: int64
              excluded: bool
              to
              {'canonical_id': Value('string'), 'excluded': Value('bool'), 'exclusion_reason': Value('string'), 'frame': Value('int64'), 'prediction_horizon_frames': Value('int64'), 'supervision_end_frame': Value('int64'), 'warning_on_frame': Value('int64')}
              because column names don't match
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1369, in compute_config_parquet_and_info_response
                  parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
                                                                        ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      builder, max_dataset_size_bytes=max_dataset_size_bytes
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 948, in stream_convert_to_parquet
                  builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
                  ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1694, in _prepare_split
                  for job_id, done, content in self._prepare_split_single(
                                               ~~~~~~~~~~~~~~~~~~~~~~~~~~^
                      gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  ):
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1880, in _prepare_split_single
                  raise DatasetGenerationError("An error occurred while generating the dataset") from e
              datasets.exceptions.DatasetGenerationError: An error occurred while generating the dataset

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

canonical_id
string
excluded
bool
exclusion_reason
string
frame
int64
prediction_horizon_frames
int64
supervision_end_frame
int64
warning_on_frame
int64
S1_A_0001
true
future_label_crosses_warning_onset
0
10
10
10
S1_A_0001
true
future_label_crosses_warning_onset
1
10
11
10
S1_A_0001
true
future_label_crosses_warning_onset
2
10
12
10
S1_A_0001
true
future_label_crosses_warning_onset
3
10
13
10
S1_A_0001
true
future_label_crosses_warning_onset
4
10
14
10
S1_A_0001
true
future_label_crosses_warning_onset
5
10
15
10
S1_A_0001
true
future_label_crosses_warning_onset
6
10
16
10
S1_A_0001
true
future_label_crosses_warning_onset
7
10
17
10
S1_A_0001
true
future_label_crosses_warning_onset
8
10
18
10
S1_A_0001
true
future_label_crosses_warning_onset
9
10
19
10
S1_A_0002
true
future_label_crosses_warning_onset
0
10
10
10
S1_A_0002
true
future_label_crosses_warning_onset
1
10
11
10
S1_A_0002
true
future_label_crosses_warning_onset
2
10
12
10
S1_A_0002
true
future_label_crosses_warning_onset
3
10
13
10
S1_A_0002
true
future_label_crosses_warning_onset
4
10
14
10
S1_A_0002
true
future_label_crosses_warning_onset
5
10
15
10
S1_A_0002
true
future_label_crosses_warning_onset
6
10
16
10
S1_A_0002
true
future_label_crosses_warning_onset
7
10
17
10
S1_A_0002
true
future_label_crosses_warning_onset
8
10
18
10
S1_A_0002
true
future_label_crosses_warning_onset
9
10
19
10
S1_A_0003
true
future_label_crosses_warning_onset
0
10
10
10
S1_A_0003
true
future_label_crosses_warning_onset
1
10
11
10
S1_A_0003
true
future_label_crosses_warning_onset
2
10
12
10
S1_A_0003
true
future_label_crosses_warning_onset
3
10
13
10
S1_A_0003
true
future_label_crosses_warning_onset
4
10
14
10
S1_A_0003
true
future_label_crosses_warning_onset
5
10
15
10
S1_A_0003
true
future_label_crosses_warning_onset
6
10
16
10
S1_A_0003
true
future_label_crosses_warning_onset
7
10
17
10
S1_A_0003
true
future_label_crosses_warning_onset
8
10
18
10
S1_A_0003
true
future_label_crosses_warning_onset
9
10
19
10
S1_A_0004
true
future_label_crosses_warning_onset
0
10
10
7
S1_A_0004
true
future_label_crosses_warning_onset
1
10
11
7
S1_A_0004
true
future_label_crosses_warning_onset
2
10
12
7
S1_A_0004
true
future_label_crosses_warning_onset
3
10
13
7
S1_A_0004
true
future_label_crosses_warning_onset
4
10
14
7
S1_A_0004
true
future_label_crosses_warning_onset
5
10
15
7
S1_A_0004
true
future_label_crosses_warning_onset
6
10
16
7
S1_A_0005
true
future_label_crosses_warning_onset
0
10
10
10
S1_A_0005
true
future_label_crosses_warning_onset
1
10
11
10
S1_A_0005
true
future_label_crosses_warning_onset
2
10
12
10
S1_A_0005
true
future_label_crosses_warning_onset
3
10
13
10
S1_A_0005
true
future_label_crosses_warning_onset
4
10
14
10
S1_A_0005
true
future_label_crosses_warning_onset
5
10
15
10
S1_A_0005
true
future_label_crosses_warning_onset
6
10
16
10
S1_A_0005
true
future_label_crosses_warning_onset
7
10
17
10
S1_A_0005
true
future_label_crosses_warning_onset
8
10
18
10
S1_A_0005
true
future_label_crosses_warning_onset
9
10
19
10
S1_A_0006
true
future_label_crosses_warning_onset
0
10
10
10
S1_A_0006
true
future_label_crosses_warning_onset
1
10
11
10
S1_A_0006
true
future_label_crosses_warning_onset
2
10
12
10
S1_A_0006
true
future_label_crosses_warning_onset
3
10
13
10
S1_A_0006
true
future_label_crosses_warning_onset
4
10
14
10
S1_A_0006
true
future_label_crosses_warning_onset
5
10
15
10
S1_A_0006
true
future_label_crosses_warning_onset
6
10
16
10
S1_A_0006
true
future_label_crosses_warning_onset
7
10
17
10
S1_A_0006
true
future_label_crosses_warning_onset
8
10
18
10
S1_A_0006
true
future_label_crosses_warning_onset
9
10
19
10
S1_A_0007
true
future_label_crosses_warning_onset
0
10
10
4
S1_A_0007
true
future_label_crosses_warning_onset
1
10
11
4
S1_A_0007
true
future_label_crosses_warning_onset
2
10
12
4
S1_A_0007
true
future_label_crosses_warning_onset
3
10
13
4
S1_A_0008
true
future_label_crosses_warning_onset
2
10
12
12
S1_A_0008
true
future_label_crosses_warning_onset
3
10
13
12
S1_A_0008
true
future_label_crosses_warning_onset
4
10
14
12
S1_A_0008
true
future_label_crosses_warning_onset
5
10
15
12
S1_A_0008
true
future_label_crosses_warning_onset
6
10
16
12
S1_A_0008
true
future_label_crosses_warning_onset
7
10
17
12
S1_A_0008
true
future_label_crosses_warning_onset
8
10
18
12
S1_A_0008
true
future_label_crosses_warning_onset
9
10
19
12
S1_A_0008
true
future_label_crosses_warning_onset
10
10
20
12
S1_A_0008
true
future_label_crosses_warning_onset
11
10
21
12
S1_A_0009
true
future_label_crosses_warning_onset
0
10
10
10
S1_A_0009
true
future_label_crosses_warning_onset
1
10
11
10
S1_A_0009
true
future_label_crosses_warning_onset
2
10
12
10
S1_A_0009
true
future_label_crosses_warning_onset
3
10
13
10
S1_A_0009
true
future_label_crosses_warning_onset
4
10
14
10
S1_A_0009
true
future_label_crosses_warning_onset
5
10
15
10
S1_A_0009
true
future_label_crosses_warning_onset
6
10
16
10
S1_A_0009
true
future_label_crosses_warning_onset
7
10
17
10
S1_A_0009
true
future_label_crosses_warning_onset
8
10
18
10
S1_A_0009
true
future_label_crosses_warning_onset
9
10
19
10
S1_A_0010
true
future_label_crosses_warning_onset
0
10
10
10
S1_A_0010
true
future_label_crosses_warning_onset
1
10
11
10
S1_A_0010
true
future_label_crosses_warning_onset
2
10
12
10
S1_A_0010
true
future_label_crosses_warning_onset
3
10
13
10
S1_A_0010
true
future_label_crosses_warning_onset
4
10
14
10
S1_A_0010
true
future_label_crosses_warning_onset
5
10
15
10
S1_A_0010
true
future_label_crosses_warning_onset
6
10
16
10
S1_A_0010
true
future_label_crosses_warning_onset
7
10
17
10
S1_A_0010
true
future_label_crosses_warning_onset
8
10
18
10
S1_A_0010
true
future_label_crosses_warning_onset
9
10
19
10
S1_A_0011
true
future_label_crosses_warning_onset
0
10
10
9
S1_A_0011
true
future_label_crosses_warning_onset
1
10
11
9
S1_A_0011
true
future_label_crosses_warning_onset
2
10
12
9
S1_A_0011
true
future_label_crosses_warning_onset
3
10
13
9
S1_A_0011
true
future_label_crosses_warning_onset
4
10
14
9
S1_A_0011
true
future_label_crosses_warning_onset
5
10
15
9
S1_A_0011
true
future_label_crosses_warning_onset
6
10
16
9
S1_A_0011
true
future_label_crosses_warning_onset
7
10
17
9
S1_A_0011
true
future_label_crosses_warning_onset
8
10
18
9
End of preview.

LANTERN

LANTERN is a benchmark for temporally grounded cooperative warnings in autonomous driving. This dataset repository releases the final open-loop training collection and the frozen closed-loop evaluation suite used by the benchmark.

Contents

  • Open-loop collection: 3,272 route sequences, 236,309 synchronized frames, and 119.8 GB of payload data.
  • Closed-loop suite: 120 hazard routes and 120 paired, context-preserving no-hazard controls across six scenario families.
  • Modalities: front RGB, measurements, cooperative annotations, 3D boxes, LiDAR, and, where available, depth, semantics, and BEV semantics. Some legacy Right-Turn Yield routes do not contain every dense auxiliary modality; openloop/manifest.jsonl records availability per route.
Internal ID Scenario family Route A Route B Frames
S1 Pedestrian Emergence 283 264 50,606
S7 Vehicle Cut-In 287 288 28,222
S8 Obstacle Reveal 227 252 36,913
S9 Right-Turn Wrong-Way 291 292 37,453
S10 Left-Turn Wrong-Way 276 267 39,916
S11 Right-Turn Yield 267 278 43,199

Route A contains the physical hazard and its aligned warning. Open-loop Route B provides ordinary-driving controls. In the closed-loop suite, every Route A has a Route B with identical route geometry, weather, and physical context; Route B disables the hazardous maneuver and warning while retaining non-hazard actors.

Download 

hf download YongshuoLiu/LANTERN --repo-type dataset --local-dir LANTERN

Open-loop data are stored as uncompressed tar shards so that millions of small sensor files can be downloaded reliably. Each route is wholly contained in one shard. Locate a route through openloop/manifest.jsonl, then extract its shard:

tar -xf openloop/shards/S1/A/S1_A-000.tar -C /path/to/output

The extracted member is routes/<canonical_id>/.... The helper tools/extract_openloop_route.py extracts a route by canonical ID.

Open-Loop Metadata

openloop/manifest.jsonl contains one row per route and records:

  • canonical route ID, scenario family, and A/B variant;
  • archive and member path;
  • frame count, byte size, and modality availability;
  • warning onset/termination and the ten-frame pre-warning exclusion window;
  • source collection tier and environment metadata.

Training code should exclude anchors listed in openloop/training_exclusions.jsonl. These anchors occur before warning onset but their future ten-waypoint target crosses into warning-conditioned behavior.

Closed-Loop Suite

The closedloop/ directory contains the route XML files, paired controls, scenario implementations, manifests, checksums, validation scripts, and runtime adapters. Validate it with:

cd closedloop
bash scripts/validate_bundle.sh
sha256sum -c SHA256SUMS

CARLA, ScenarioRunner, Bench2Drive, and model-specific agents are external dependencies and are not redistributed in this dataset repository. See closedloop/README.md for example commands.

Integrity and Scope

  • The public manifests contain no machine-specific absolute paths.
  • Test videos, evaluator logs, temporary runtime overlays, rejected routes, and Python caches are excluded.
  • Tar shards are not compressed because most image and sensor payloads are already compressed.
  • This release is simulator-derived. Users must comply with the licenses of CARLA and any external software used to run the benchmark.
Downloads last month
74
Total file size:
127 GB