Hugging Face's logo

AutowareFoundation
/
latent_drive

Robotics
TensorRT
ONNX
autoware
ros2
autonomous-driving
planning
end-to-end
e2e
trajectory-generation
camera
video
world-model
latent-world-model
v-jepa
jepa
latentdrive
Model card Files
xet
 Community
kalfazed commited on
Commit
83f7eb0
·
verified ·
1 Parent(s): 145fbb5

Replace the lane change with a right turn in the scene replays

Browse files
Files changed (4) hide show
  1. README.md +3 -3
  2. SHA256SUMS +2 -2
  3. media/latent_drive_scenes.png +2 -2
  4. replay.mp4 +2 -2
README.md CHANGED
@@ -26,7 +26,7 @@ route and the ego motion into a 4 s ego trajectory, at 10 Hz. A V-JEPA ViT-L vid
26
  a small planning head reads the encoder's features together with the route and the ego state.
27
 
28
  <p align="center">
29
- <img src="media/latent_drive_scenes.png" width="100%" alt="Six scenes with the three-camera model's plan projected on the front camera: following a lane at 40 km/h, changing into the right lane, a left turn at an intersection, starting on green and turning left, slowing down for a red light, and starting on green">
30
  </p>
31
 
32
  Two models are published here, sharing the encoder, the route input and the output:
@@ -55,7 +55,7 @@ Two models are published here, sharing the encoder, the route input and the outp
55
  | `camfront3-rptw-lane2-e1.onnx` | The three-camera model (1.25 GB) |
56
  | `camfront-rptw-lane2-e1.onnx` | The one-camera model (1.24 GB) |
57
  | `media/` | The scenes on this page |
58
- | `replay.mp4` | Open-loop replays of six scenes of one recorded drive, the three-camera model's plan drawn on the front camera; this page's video preview |
59
  | `deploy_metadata.yaml` | Deployment metadata recording the artifact version of this repository |
60
  | `SHA256SUMS` | Checksums of the files above |
61
 
@@ -143,7 +143,7 @@ validation data and are not published.
143
  - **Known behaviours.** Where the route doubles back (a junction entry whose centerline turns back), near route
144
  points can land behind or beside the car and the plan cuts in; `autoware_latent_drive` drops route vertices
145
  that turn by more than 120°. The models creep slightly while the car is stopped.
146
- - **Evaluation.** The video preview and the scenes on this page are open-loop replays of a recorded drive
147
  (Tokyo, Ariake, 2026-10-23), made with `samples/rosbag/run_replay.sh`: the plan is drawn on the camera
148
  frames, not executed. The ribbon runs green (now) to red (+4 s).
149
 
 
26
  a small planning head reads the encoder's features together with the route and the ego state.
27
 
28
  <p align="center">
29
+ <img src="media/latent_drive_scenes.png" width="100%" alt="Six scenes with the three-camera model's plan projected on the front camera: following a lane at 40 km/h, turning right at an intersection after yielding to oncoming traffic, a left turn at an intersection, starting on green and turning left, slowing down for a red light, and starting on green">
30
  </p>
31
 
32
  Two models are published here, sharing the encoder, the route input and the output:
 
55
  | `camfront3-rptw-lane2-e1.onnx` | The three-camera model (1.25 GB) |
56
  | `camfront-rptw-lane2-e1.onnx` | The one-camera model (1.24 GB) |
57
  | `media/` | The scenes on this page |
58
+ | `replay.mp4` | Open-loop replays of six scenes of two recorded drives, the three-camera model's plan drawn on the front camera; this page's video preview |
59
  | `deploy_metadata.yaml` | Deployment metadata recording the artifact version of this repository |
60
  | `SHA256SUMS` | Checksums of the files above |
61
 
 
143
  - **Known behaviours.** Where the route doubles back (a junction entry whose centerline turns back), near route
144
  points can land behind or beside the car and the plan cuts in; `autoware_latent_drive` drops route vertices
145
  that turn by more than 120°. The models creep slightly while the car is stopped.
146
+ - **Evaluation.** The video preview and the scenes on this page are open-loop replays of two recorded drives
147
  (Tokyo, Ariake, 2026-10-23), made with `samples/rosbag/run_replay.sh`: the plan is drawn on the camera
148
  frames, not executed. The ribbon runs green (now) to red (+4 s).
149
 
SHA256SUMS CHANGED
@@ -2,5 +2,5 @@
2
  8aa2f982fd0b07aac305d75a6522236c9b307ede7ca09223bb92381035c21659 camfront-rptw-lane2-e1.onnx
3
  6bd2eba88475af5e5a823eb39b6994e1c45a820049e73500f40a114dd0f3fb03 deploy_metadata.yaml
4
  c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4 LICENSE
5
- 0747053330b87ed8629859b4712a9cb94913c3d0133b0866eb6a06f5147e499f media/latent_drive_scenes.png
6
- 1fc9088fbfad826d4061c19fdb15803bf38320e107ef2b32bd0f5513362ceb2f replay.mp4
 
2
  8aa2f982fd0b07aac305d75a6522236c9b307ede7ca09223bb92381035c21659 camfront-rptw-lane2-e1.onnx
3
  6bd2eba88475af5e5a823eb39b6994e1c45a820049e73500f40a114dd0f3fb03 deploy_metadata.yaml
4
  c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4 LICENSE
5
+ 2dfcb26d20851b243d19aa689a15b371c13c655e7e86598197c9f19329772444 media/latent_drive_scenes.png
6
+ b9bb481aa6084343cc43308dd192512834a36d3b4746084cd316a28e14fc9060 replay.mp4
media/latent_drive_scenes.png CHANGED

Git LFS Details

  • SHA256: 0747053330b87ed8629859b4712a9cb94913c3d0133b0866eb6a06f5147e499f
  • Pointer size: 132 Bytes
  • Size of remote file: 1.32 MB

Git LFS Details

  • SHA256: 2dfcb26d20851b243d19aa689a15b371c13c655e7e86598197c9f19329772444
  • Pointer size: 132 Bytes
  • Size of remote file: 1.35 MB
replay.mp4 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:1fc9088fbfad826d4061c19fdb15803bf38320e107ef2b32bd0f5513362ceb2f
3
- size 27390883
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b9bb481aa6084343cc43308dd192512834a36d3b4746084cd316a28e14fc9060
3
+ size 27499977