Hugging Face's logo

AutowareFoundation
/
diffusion_planner

Robotics
TensorRT
ONNX
autoware
ros2
autonomous-driving
planning
trajectory-generation
diffusion
Model card Files
xet
 Community
xmfcx commited on
Commit
28ce647
·
verified ·
1 Parent(s): 86864fb

feat: add diffusion_planner v3.0 artifacts (from awf.ml.dev.web.auto/planning/models/diffusion_planner/v3.0)

Browse files
Files changed (5) hide show
  1. .gitignore +4 -0
  2. README.md +163 -0
  3. deploy_metadata.yaml +1 -0
  4. diffusion_planner.onnx +3 -0
  5. diffusion_planner.param.json +880 -0
.gitignore ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ # Auto-generated TensorRT artifacts, built locally by Autoware from the ONNX
2
+ # files. They are environment-specific (GPU arch + TensorRT version) and must
3
+ # not be committed to this repo.
4
+ *.engine
README.md ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ pipeline_tag: robotics
4
+ tags:
5
+ - autoware
6
+ - ros2
7
+ - autonomous-driving
8
+ - planning
9
+ - trajectory-generation
10
+ - diffusion
11
+ - tensorrt
12
+ - onnx
13
+ ---
14
+
15
+ # Diffusion Planner for Autoware (`diffusion_planner`)
16
+
17
+ Trajectory generation models for autonomous driving, used by the
18
+ [`autoware_diffusion_planner`](https://github.com/autowarefoundation/autoware_universe/tree/main/planning/autoware_diffusion_planner)
19
+ node in [Autoware](https://github.com/autowarefoundation/autoware).
20
+
21
+ The models follow the **Diffusion Planner** [1] architecture from "Diffusion-Based Planning for Autonomous
22
+ Driving with Flexible Guidance" (Zheng et al.). The node generates smooth, feasible, and safe ego trajectories
23
+ by considering dynamic and static obstacles, vehicle kinematics, Lanelet2 map context, the route, and traffic
24
+ signals with speed limits. Models are distributed as ONNX and run with TensorRT (default) or ONNX Runtime.
25
+
26
+ ## Model overview
27
+
28
+ | | |
29
+ | --- | --- |
30
+ | Task | Ego trajectory generation (plus candidate trajectories, predicted objects, and turn indicator commands) |
31
+ | Architecture | Diffusion Planner (diffusion-based trajectory generation) |
32
+ | Runtime | TensorRT (default, FP32) or ONNX Runtime (CPU, CUDA, or TensorRT execution provider) via the `autoware_diffusion_planner` ROS 2 node |
33
+ | Format | ONNX (Autoware builds the TensorRT engine locally on first launch) |
34
+ | License | Apache-2.0 |
35
+
36
+ ## Versions in this repository
37
+
38
+ This is a multi-version repository. Each git tag contains exactly that version's file set at the repository
39
+ root; `main` always points to the newest tag (currently `v5.0`). Consumers must pin `--revision <tag>` and
40
+ never rely on `main`. Autoware installs each version side by side under
41
+ `~/autoware_data/ml_models/diffusion_planner/<tag>/`.
42
+
43
+ | Tag | Release date | Files | Notes (from the package's model version history) |
44
+ | --- | --- | --- | --- |
45
+ | `v3.0` | 2026/01/09 | `diffusion_planner.onnx`, `diffusion_planner.param.json`, `deploy_metadata.yaml` | Added `TURN_INDICATOR_OUTPUT_KEEP` output; supervised fine-tuning (SFT) with carefully filtered data; encoder layers increased from 3 to 6 |
46
+ | `v3.1` | 2026/03/05 | `diffusion_planner.onnx`, `diffusion_planner.param.json`, `deploy_metadata.yaml` | ONNX-simplified model for faster TensorRT engine build and reduced GPU memory; same weights as v3.0 (no retraining) |
47
+ | `v4.0` | 2026/03/23 | `diffusion_planner.onnx`, `diffusion_planner.param.json`, `deploy_metadata.yaml` | Added `delay` input for Real-Time Chunking (RTC); one-hot type encoding for polygons and line strings; `NUM_LINE_STRINGS` increased from 10 to 60; line string resampling |
48
+ | `v5.0` | not documented | v4.0 set plus `diffusion_planner_encoder.onnx`, `diffusion_planner_decoder.onnx`, `diffusion_planner_turn_indicator.onnx` | Adds separate encoder, decoder, and turn indicator sub-models used by the node's `multi_step` mode (configurable DPM solver steps) |
49
+
50
+ The model versioning scheme uses major and minor numbers: the major version changes when model inputs/outputs
51
+ or architecture change (models with a different major version are not compatible with the ROS node), and the
52
+ minor version changes when only the weights are updated. The package also documents older model versions
53
+ (0.1, 1.0, 2.0) that are not compatible with the current node; they are not distributed here.
54
+
55
+ ## Files
56
+
57
+ Files at `main` / `v5.0` (earlier tags contain `diffusion_planner.onnx`, `diffusion_planner.param.json`,
58
+ and `deploy_metadata.yaml`):
59
+
60
+ | File | Description |
61
+ | --- | --- |
62
+ | `diffusion_planner.onnx` | Single-step model (full network in one ONNX graph) |
63
+ | `diffusion_planner_encoder.onnx` | Encoder sub-model for multi-step inference |
64
+ | `diffusion_planner_decoder.onnx` | Decoder sub-model for multi-step inference (run for a configurable number of DPM solver steps) |
65
+ | `diffusion_planner_turn_indicator.onnx` | Turn indicator sub-model |
66
+ | `diffusion_planner.param.json` | Model parameters consumed by the node alongside the ONNX files |
67
+ | `deploy_metadata.yaml` | Deployment metadata recording the artifact version of this repository |
68
+
69
+ > **TensorRT engines are not distributed here.** TensorRT engines are specific to the GPU architecture and
70
+ > TensorRT version they are built on and are not portable, so Autoware builds them locally from the ONNX files
71
+ > on first launch (or via `build_only:=true`).
72
+
73
+ ## Inputs and outputs (as used by the node)
74
+
75
+ **Inputs**
76
+
77
+ | Topic | Message Type | Description |
78
+ | --- | --- | --- |
79
+ | `~/input/odometry` | `nav_msgs/msg/Odometry` | Ego vehicle odometry |
80
+ | `~/input/acceleration` | `geometry_msgs/msg/AccelWithCovarianceStamped` | Ego acceleration |
81
+ | `~/input/tracked_objects` | `autoware_perception_msgs/msg/TrackedObjects` | Detected dynamic objects |
82
+ | `~/input/traffic_signals` | `autoware_perception_msgs/msg/TrafficLightGroupArray` | Traffic light states |
83
+ | `~/input/vector_map` | `autoware_map_msgs/msg/LaneletMapBin` | Lanelet2 map |
84
+ | `~/input/route` | `autoware_planning_msgs/msg/LaneletRoute` | Route information |
85
+ | `~/input/turn_indicators` | `autoware_vehicle_msgs/msg/TurnIndicatorsReport` | Turn indicator information |
86
+
87
+ **Outputs**
88
+
89
+ | Topic | Message Type | Description |
90
+ | --- | --- | --- |
91
+ | `~/output/trajectory` | `autoware_planning_msgs/msg/Trajectory` | Planned trajectory for the ego vehicle |
92
+ | `~/output/trajectories` | `autoware_internal_planning_msgs/msg/CandidateTrajectories` | Multiple candidate trajectories |
93
+ | `~/output/predicted_objects` | `autoware_perception_msgs/msg/PredictedObjects` | Predicted future states of dynamic objects |
94
+ | `~/output/turn_indicators` | `autoware_vehicle_msgs/msg/TurnIndicatorsCommand` | Planned turn indicator command |
95
+ | `~/output/debug/traffic_signal` | `autoware_perception_msgs/msg/TrafficLightGroup` | First traffic light on route (ego forward) |
96
+ | `~/debug/lane_marker` | `visualization_msgs/msg/MarkerArray` | Lane debug markers |
97
+ | `~/debug/route_marker` | `visualization_msgs/msg/MarkerArray` | Route debug markers |
98
+
99
+ ## Usage in Autoware
100
+
101
+ Autoware downloads these artifacts to `~/autoware_data/ml_models/diffusion_planner/<tag>/` (see
102
+ [Download artifacts](https://github.com/autowarefoundation/autoware/blob/main/ansible/roles/artifacts/README.md#download-artifacts)).
103
+ The node's config (`config/diffusion_planner.param.yaml`) points at one version directory, currently `v5.0`.
104
+
105
+ Launch the planning simulator with the diffusion planner selected:
106
+
107
+ ```bash
108
+ ros2 launch autoware_launch planning_simulator.launch.xml \
109
+ map_path:=/path/to/your/map \
110
+ vehicle_model:=sample_vehicle \
111
+ sensor_model:=sample_sensor_kit \
112
+ planning_setting:=diffusion_planner
113
+ ```
114
+
115
+ `planning_setting:=diffusion_planner` swaps the trajectory generator, the planning validator input topic, and
116
+ the diagnostics graph, so no additional launch-file edits are required. The node's own launch file
117
+ (`diffusion_planner.launch.xml`) accepts `build_only:=true` to shut the node down after model initialization,
118
+ which can be used to pre-build the TensorRT engine.
119
+ See the [package README](https://github.com/autowarefoundation/autoware_universe/tree/main/planning/autoware_diffusion_planner)
120
+ for the full parameter reference.
121
+
122
+ ## Provenance
123
+
124
+ These artifacts were previously hosted at
125
+ `https://awf.ml.dev.web.auto/planning/models/diffusion_planner/<version>/` for versions `v3.0` through
126
+ `v5.0`. The HF tags map one-to-one to those source version directories, so traceability to the original
127
+ hosting is preserved.
128
+
129
+ ## Training
130
+
131
+ - Training fork (used to train these models): <https://github.com/tier4/Diffusion-Planner>
132
+ - Original implementation: <https://github.com/ZhengYinan-AIR/Diffusion-Planner>
133
+ - Paper: "Diffusion-Based Planning for Autonomous Driving with Flexible Guidance", arXiv:2501.15564
134
+
135
+ The models were trained by TIER IV. The package's model version history records that training used TIER IV
136
+ synthetic and real driving data, with supervised fine-tuning on carefully filtered data introduced in v3.0; the
137
+ exact dataset composition per version is not publicly documented.
138
+
139
+ ## Limitations
140
+
141
+ - Models with a major version different from the one the ROS node expects are not compatible with the node.
142
+ - The node is aimed at the proposed [Autoware new planning framework](https://github.com/tier4/new_planning_framework).
143
+ - Training data is TIER IV internal; performance on other vehicle platforms and operational domains is not
144
+ publicly characterized.
145
+
146
+ ## Citation
147
+
148
+ ```bibtex
149
+ @article{zheng2025diffusionplanner,
150
+ title = {Diffusion-Based Planning for Autonomous Driving with Flexible Guidance},
151
+ author = {Zheng, Yinan and others},
152
+ journal = {arXiv preprint arXiv:2501.15564},
153
+ year = {2025}
154
+ }
155
+ ```
156
+
157
+ ## References
158
+
159
+ - [1] Zheng et al., "Diffusion-Based Planning for Autonomous Driving with Flexible Guidance", arXiv:2501.15564, 2025.
160
+ - Original implementation: <https://github.com/ZhengYinan-AIR/Diffusion-Planner>
161
+ - Training fork: <https://github.com/tier4/Diffusion-Planner>
162
+ - Consuming package: <https://github.com/autowarefoundation/autoware_universe/tree/main/planning/autoware_diffusion_planner>
163
+ - Autoware: <https://github.com/autowarefoundation/autoware>
deploy_metadata.yaml ADDED
@@ -0,0 +1 @@
 
 
1
+ version: v3.0
diffusion_planner.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3026918b55869b02ea561c1e1b9fed05c34092294c035d743f1d1cb756f4abcc
3
+ size 61521099
diffusion_planner.param.json ADDED
@@ -0,0 +1,880 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "future_len": 80,
3
+ "time_len": 31,
4
+ "agent_state_dim": 11,
5
+ "agent_num": 32,
6
+ "static_objects_state_dim": 10,
7
+ "static_objects_num": 5,
8
+ "lane_num": 140,
9
+ "lane_len": 20,
10
+ "route_num": 25,
11
+ "route_len": 20,
12
+ "polygon_num": 10,
13
+ "polygon_len": 40,
14
+ "line_string_num": 10,
15
+ "line_string_len": 20,
16
+ "use_data_augment": true,
17
+ "augment_prob": 0.5,
18
+ "num_workers": 4,
19
+ "pin_mem": true,
20
+ "seed": 3407,
21
+ "train_epochs": 200,
22
+ "early_stop_tolerance": 100,
23
+ "batch_size": 1024,
24
+ "save_utd": 10,
25
+ "learning_rate": 2e-05,
26
+ "warm_up_epoch": 5,
27
+ "encoder_drop_path_rate": 0.1,
28
+ "decoder_drop_path_rate": 0.1,
29
+ "use_ego_history": true,
30
+ "ego_history_dropout_rate": 0.6,
31
+ "use_turn_indicators": true,
32
+ "coeff_position_lat_loss": 1.0,
33
+ "coeff_position_lon_loss": 1.0,
34
+ "coeff_heading_l2_loss": 1.0,
35
+ "coeff_velocity": 1.0,
36
+ "coeff_timestep": [
37
+ 1.0,
38
+ 1.0,
39
+ 1.0,
40
+ 1.0
41
+ ],
42
+ "alpha_planning_loss": 1.0,
43
+ "device": "cuda",
44
+ "use_ema": true,
45
+ "encoder_mixer_depth": 6,
46
+ "encoder_fusion_depth": 6,
47
+ "decoder_depth": 3,
48
+ "num_heads": 8,
49
+ "hidden_dim": 256,
50
+ "diffusion_model_type": "x_start",
51
+ "predicted_neighbor_num": 32,
52
+ "use_wandb": true,
53
+ "notes": "",
54
+ "ddp": true,
55
+ "state_normalizer": {
56
+ "mean": [
57
+ [
58
+ [
59
+ 10,
60
+ 0,
61
+ 0,
62
+ 0
63
+ ]
64
+ ],
65
+ [
66
+ [
67
+ 10,
68
+ 0,
69
+ 0,
70
+ 0
71
+ ]
72
+ ],
73
+ [
74
+ [
75
+ 10,
76
+ 0,
77
+ 0,
78
+ 0
79
+ ]
80
+ ],
81
+ [
82
+ [
83
+ 10,
84
+ 0,
85
+ 0,
86
+ 0
87
+ ]
88
+ ],
89
+ [
90
+ [
91
+ 10,
92
+ 0,
93
+ 0,
94
+ 0
95
+ ]
96
+ ],
97
+ [
98
+ [
99
+ 10,
100
+ 0,
101
+ 0,
102
+ 0
103
+ ]
104
+ ],
105
+ [
106
+ [
107
+ 10,
108
+ 0,
109
+ 0,
110
+ 0
111
+ ]
112
+ ],
113
+ [
114
+ [
115
+ 10,
116
+ 0,
117
+ 0,
118
+ 0
119
+ ]
120
+ ],
121
+ [
122
+ [
123
+ 10,
124
+ 0,
125
+ 0,
126
+ 0
127
+ ]
128
+ ],
129
+ [
130
+ [
131
+ 10,
132
+ 0,
133
+ 0,
134
+ 0
135
+ ]
136
+ ],
137
+ [
138
+ [
139
+ 10,
140
+ 0,
141
+ 0,
142
+ 0
143
+ ]
144
+ ],
145
+ [
146
+ [
147
+ 10,
148
+ 0,
149
+ 0,
150
+ 0
151
+ ]
152
+ ],
153
+ [
154
+ [
155
+ 10,
156
+ 0,
157
+ 0,
158
+ 0
159
+ ]
160
+ ],
161
+ [
162
+ [
163
+ 10,
164
+ 0,
165
+ 0,
166
+ 0
167
+ ]
168
+ ],
169
+ [
170
+ [
171
+ 10,
172
+ 0,
173
+ 0,
174
+ 0
175
+ ]
176
+ ],
177
+ [
178
+ [
179
+ 10,
180
+ 0,
181
+ 0,
182
+ 0
183
+ ]
184
+ ],
185
+ [
186
+ [
187
+ 10,
188
+ 0,
189
+ 0,
190
+ 0
191
+ ]
192
+ ],
193
+ [
194
+ [
195
+ 10,
196
+ 0,
197
+ 0,
198
+ 0
199
+ ]
200
+ ],
201
+ [
202
+ [
203
+ 10,
204
+ 0,
205
+ 0,
206
+ 0
207
+ ]
208
+ ],
209
+ [
210
+ [
211
+ 10,
212
+ 0,
213
+ 0,
214
+ 0
215
+ ]
216
+ ],
217
+ [
218
+ [
219
+ 10,
220
+ 0,
221
+ 0,
222
+ 0
223
+ ]
224
+ ],
225
+ [
226
+ [
227
+ 10,
228
+ 0,
229
+ 0,
230
+ 0
231
+ ]
232
+ ],
233
+ [
234
+ [
235
+ 10,
236
+ 0,
237
+ 0,
238
+ 0
239
+ ]
240
+ ],
241
+ [
242
+ [
243
+ 10,
244
+ 0,
245
+ 0,
246
+ 0
247
+ ]
248
+ ],
249
+ [
250
+ [
251
+ 10,
252
+ 0,
253
+ 0,
254
+ 0
255
+ ]
256
+ ],
257
+ [
258
+ [
259
+ 10,
260
+ 0,
261
+ 0,
262
+ 0
263
+ ]
264
+ ],
265
+ [
266
+ [
267
+ 10,
268
+ 0,
269
+ 0,
270
+ 0
271
+ ]
272
+ ],
273
+ [
274
+ [
275
+ 10,
276
+ 0,
277
+ 0,
278
+ 0
279
+ ]
280
+ ],
281
+ [
282
+ [
283
+ 10,
284
+ 0,
285
+ 0,
286
+ 0
287
+ ]
288
+ ],
289
+ [
290
+ [
291
+ 10,
292
+ 0,
293
+ 0,
294
+ 0
295
+ ]
296
+ ],
297
+ [
298
+ [
299
+ 10,
300
+ 0,
301
+ 0,
302
+ 0
303
+ ]
304
+ ],
305
+ [
306
+ [
307
+ 10,
308
+ 0,
309
+ 0,
310
+ 0
311
+ ]
312
+ ],
313
+ [
314
+ [
315
+ 10,
316
+ 0,
317
+ 0,
318
+ 0
319
+ ]
320
+ ]
321
+ ],
322
+ "std": [
323
+ [
324
+ [
325
+ 20,
326
+ 20,
327
+ 1,
328
+ 1
329
+ ]
330
+ ],
331
+ [
332
+ [
333
+ 20,
334
+ 20,
335
+ 1,
336
+ 1
337
+ ]
338
+ ],
339
+ [
340
+ [
341
+ 20,
342
+ 20,
343
+ 1,
344
+ 1
345
+ ]
346
+ ],
347
+ [
348
+ [
349
+ 20,
350
+ 20,
351
+ 1,
352
+ 1
353
+ ]
354
+ ],
355
+ [
356
+ [
357
+ 20,
358
+ 20,
359
+ 1,
360
+ 1
361
+ ]
362
+ ],
363
+ [
364
+ [
365
+ 20,
366
+ 20,
367
+ 1,
368
+ 1
369
+ ]
370
+ ],
371
+ [
372
+ [
373
+ 20,
374
+ 20,
375
+ 1,
376
+ 1
377
+ ]
378
+ ],
379
+ [
380
+ [
381
+ 20,
382
+ 20,
383
+ 1,
384
+ 1
385
+ ]
386
+ ],
387
+ [
388
+ [
389
+ 20,
390
+ 20,
391
+ 1,
392
+ 1
393
+ ]
394
+ ],
395
+ [
396
+ [
397
+ 20,
398
+ 20,
399
+ 1,
400
+ 1
401
+ ]
402
+ ],
403
+ [
404
+ [
405
+ 20,
406
+ 20,
407
+ 1,
408
+ 1
409
+ ]
410
+ ],
411
+ [
412
+ [
413
+ 20,
414
+ 20,
415
+ 1,
416
+ 1
417
+ ]
418
+ ],
419
+ [
420
+ [
421
+ 20,
422
+ 20,
423
+ 1,
424
+ 1
425
+ ]
426
+ ],
427
+ [
428
+ [
429
+ 20,
430
+ 20,
431
+ 1,
432
+ 1
433
+ ]
434
+ ],
435
+ [
436
+ [
437
+ 20,
438
+ 20,
439
+ 1,
440
+ 1
441
+ ]
442
+ ],
443
+ [
444
+ [
445
+ 20,
446
+ 20,
447
+ 1,
448
+ 1
449
+ ]
450
+ ],
451
+ [
452
+ [
453
+ 20,
454
+ 20,
455
+ 1,
456
+ 1
457
+ ]
458
+ ],
459
+ [
460
+ [
461
+ 20,
462
+ 20,
463
+ 1,
464
+ 1
465
+ ]
466
+ ],
467
+ [
468
+ [
469
+ 20,
470
+ 20,
471
+ 1,
472
+ 1
473
+ ]
474
+ ],
475
+ [
476
+ [
477
+ 20,
478
+ 20,
479
+ 1,
480
+ 1
481
+ ]
482
+ ],
483
+ [
484
+ [
485
+ 20,
486
+ 20,
487
+ 1,
488
+ 1
489
+ ]
490
+ ],
491
+ [
492
+ [
493
+ 20,
494
+ 20,
495
+ 1,
496
+ 1
497
+ ]
498
+ ],
499
+ [
500
+ [
501
+ 20,
502
+ 20,
503
+ 1,
504
+ 1
505
+ ]
506
+ ],
507
+ [
508
+ [
509
+ 20,
510
+ 20,
511
+ 1,
512
+ 1
513
+ ]
514
+ ],
515
+ [
516
+ [
517
+ 20,
518
+ 20,
519
+ 1,
520
+ 1
521
+ ]
522
+ ],
523
+ [
524
+ [
525
+ 20,
526
+ 20,
527
+ 1,
528
+ 1
529
+ ]
530
+ ],
531
+ [
532
+ [
533
+ 20,
534
+ 20,
535
+ 1,
536
+ 1
537
+ ]
538
+ ],
539
+ [
540
+ [
541
+ 20,
542
+ 20,
543
+ 1,
544
+ 1
545
+ ]
546
+ ],
547
+ [
548
+ [
549
+ 20,
550
+ 20,
551
+ 1,
552
+ 1
553
+ ]
554
+ ],
555
+ [
556
+ [
557
+ 20,
558
+ 20,
559
+ 1,
560
+ 1
561
+ ]
562
+ ],
563
+ [
564
+ [
565
+ 20,
566
+ 20,
567
+ 1,
568
+ 1
569
+ ]
570
+ ],
571
+ [
572
+ [
573
+ 20,
574
+ 20,
575
+ 1,
576
+ 1
577
+ ]
578
+ ],
579
+ [
580
+ [
581
+ 20,
582
+ 20,
583
+ 1,
584
+ 1
585
+ ]
586
+ ]
587
+ ]
588
+ },
589
+ "observation_normalizer": {
590
+ "ego_agent_past": {
591
+ "mean": [
592
+ 10.0,
593
+ 0.0,
594
+ 0.0,
595
+ 0.0
596
+ ],
597
+ "std": [
598
+ 20.0,
599
+ 20.0,
600
+ 1.0,
601
+ 1.0
602
+ ]
603
+ },
604
+ "ego_current_state": {
605
+ "mean": [
606
+ 10.0,
607
+ 0.0,
608
+ 0.0,
609
+ 0.0,
610
+ 0.0,
611
+ 0.0,
612
+ 0.0,
613
+ 0.0,
614
+ 0.0,
615
+ 0.0
616
+ ],
617
+ "std": [
618
+ 20.0,
619
+ 20.0,
620
+ 1.0,
621
+ 1.0,
622
+ 20.0,
623
+ 20.0,
624
+ 20.0,
625
+ 20.0,
626
+ 1.0,
627
+ 1.0
628
+ ]
629
+ },
630
+ "neighbor_agents_past": {
631
+ "mean": [
632
+ 10.0,
633
+ 0.0,
634
+ 0.0,
635
+ 0.0,
636
+ 0.0,
637
+ 0.0,
638
+ 0.0,
639
+ 0.0,
640
+ 0.0,
641
+ 0.0,
642
+ 0.0
643
+ ],
644
+ "std": [
645
+ 20.0,
646
+ 20.0,
647
+ 1.0,
648
+ 1.0,
649
+ 20.0,
650
+ 20.0,
651
+ 20.0,
652
+ 20.0,
653
+ 1.0,
654
+ 1.0,
655
+ 1.0
656
+ ]
657
+ },
658
+ "static_objects": {
659
+ "mean": [
660
+ 10.0,
661
+ 0.0,
662
+ 0.0,
663
+ 0.0,
664
+ 0.0,
665
+ 0.0,
666
+ 0.0,
667
+ 0.0,
668
+ 0.0,
669
+ 0.0
670
+ ],
671
+ "std": [
672
+ 20.0,
673
+ 20.0,
674
+ 1.0,
675
+ 1.0,
676
+ 20.0,
677
+ 20.0,
678
+ 1.0,
679
+ 1.0,
680
+ 1.0,
681
+ 1.0
682
+ ]
683
+ },
684
+ "lanes": {
685
+ "mean": [
686
+ 10.0,
687
+ 0.0,
688
+ 0.0,
689
+ 0.0,
690
+ 0.0,
691
+ 0.0,
692
+ 0.0,
693
+ 0.0,
694
+ 0.0,
695
+ 0.0,
696
+ 0.0,
697
+ 0.0,
698
+ 0.0,
699
+ 0.0,
700
+ 0.0,
701
+ 0.0,
702
+ 0.0,
703
+ 0.0,
704
+ 0.0,
705
+ 0.0,
706
+ 0.0,
707
+ 0.0,
708
+ 0.0,
709
+ 0.0,
710
+ 0.0,
711
+ 0.0,
712
+ 0.0,
713
+ 0.0,
714
+ 0.0,
715
+ 0.0,
716
+ 0.0,
717
+ 0.0,
718
+ 0.0
719
+ ],
720
+ "std": [
721
+ 20.0,
722
+ 20.0,
723
+ 20.0,
724
+ 20.0,
725
+ 20.0,
726
+ 20.0,
727
+ 20.0,
728
+ 20.0,
729
+ 1.0,
730
+ 1.0,
731
+ 1.0,
732
+ 1.0,
733
+ 1.0,
734
+ 1.0,
735
+ 1.0,
736
+ 1.0,
737
+ 1.0,
738
+ 1.0,
739
+ 1.0,
740
+ 1.0,
741
+ 1.0,
742
+ 1.0,
743
+ 1.0,
744
+ 1.0,
745
+ 1.0,
746
+ 1.0,
747
+ 1.0,
748
+ 1.0,
749
+ 1.0,
750
+ 1.0,
751
+ 1.0,
752
+ 1.0,
753
+ 1.0
754
+ ]
755
+ },
756
+ "lanes_speed_limit": {
757
+ "mean": [
758
+ 0.0
759
+ ],
760
+ "std": [
761
+ 20.0
762
+ ]
763
+ },
764
+ "route_lanes": {
765
+ "mean": [
766
+ 10.0,
767
+ 0.0,
768
+ 0.0,
769
+ 0.0,
770
+ 0.0,
771
+ 0.0,
772
+ 0.0,
773
+ 0.0,
774
+ 0.0,
775
+ 0.0,
776
+ 0.0,
777
+ 0.0,
778
+ 0.0,
779
+ 0.0,
780
+ 0.0,
781
+ 0.0,
782
+ 0.0,
783
+ 0.0,
784
+ 0.0,
785
+ 0.0,
786
+ 0.0,
787
+ 0.0,
788
+ 0.0,
789
+ 0.0,
790
+ 0.0,
791
+ 0.0,
792
+ 0.0,
793
+ 0.0,
794
+ 0.0,
795
+ 0.0,
796
+ 0.0,
797
+ 0.0,
798
+ 0.0
799
+ ],
800
+ "std": [
801
+ 20.0,
802
+ 20.0,
803
+ 20.0,
804
+ 20.0,
805
+ 20.0,
806
+ 20.0,
807
+ 20.0,
808
+ 20.0,
809
+ 1.0,
810
+ 1.0,
811
+ 1.0,
812
+ 1.0,
813
+ 1.0,
814
+ 1.0,
815
+ 1.0,
816
+ 1.0,
817
+ 1.0,
818
+ 1.0,
819
+ 1.0,
820
+ 1.0,
821
+ 1.0,
822
+ 1.0,
823
+ 1.0,
824
+ 1.0,
825
+ 1.0,
826
+ 1.0,
827
+ 1.0,
828
+ 1.0,
829
+ 1.0,
830
+ 1.0,
831
+ 1.0,
832
+ 1.0,
833
+ 1.0
834
+ ]
835
+ },
836
+ "polygons": {
837
+ "mean": [
838
+ 10.0,
839
+ 0.0
840
+ ],
841
+ "std": [
842
+ 20.0,
843
+ 20.0
844
+ ]
845
+ },
846
+ "line_strings": {
847
+ "mean": [
848
+ 10.0,
849
+ 0.0
850
+ ],
851
+ "std": [
852
+ 20.0,
853
+ 20.0
854
+ ]
855
+ },
856
+ "route_lanes_speed_limit": {
857
+ "mean": [
858
+ 0.0
859
+ ],
860
+ "std": [
861
+ 20.0
862
+ ]
863
+ },
864
+ "goal_pose": {
865
+ "mean": [
866
+ 10.0,
867
+ 0.0,
868
+ 0.0,
869
+ 0.0
870
+ ],
871
+ "std": [
872
+ 20.0,
873
+ 20.0,
874
+ 1.0,
875
+ 1.0
876
+ ]
877
+ }
878
+ },
879
+ "major_version": 3
880
+ }