formatter

omniverse/extension/omni.viplanner/omni/viplanner/collectors/terrain_analysis.py

@@ -401,7 +401,9 @@ def _point_filter_wall_closeness(
         heights = heights[without_wall]
         return ray_origins, heights
 
-    def _point_filter_semantic_cost(self, ray_origins: torch.Tensor, heights: torch.Tensor) -> tuple[torch.Tensor, torch.Tensor]:
+    def _point_filter_semantic_cost(
+        self, ray_origins: torch.Tensor, heights: torch.Tensor
+    ) -> tuple[torch.Tensor, torch.Tensor]:
         # raycast vertically down and get the corresponding face id
         ray_directions = torch.zeros((ray_origins.shape[0], 3), dtype=torch.float32, device=self.device)
         ray_directions[:, 2] = -1.0

omniverse/extension/omni.viplanner/omni/viplanner/collectors/viewpoint_sampling.py

@@ -14,8 +14,8 @@
 
 import cv2
 import numpy as np
-import torch
 import omni.isaac.lab.utils.math as math_utils
+import torch
 from omni.isaac.lab.markers import VisualizationMarkers
 from omni.isaac.lab.markers.config import GREEN_ARROW_X_MARKER_CFG
 from omni.isaac.lab.scene import InteractiveScene
@@ -78,7 +78,9 @@ def sample_viewpoints(self, nbr_viewpoints: int, seed: int = 1) -> torch.Tensor:
         while sample_locations_count < nbr_viewpoints:
             # get samples
             sample_idx = self.terrain_analyser.samples[:, 0] == curr_point_idx
-            sample_idx_select = torch.randperm(sample_idx.sum())[:min(nbr_samples_per_point, nbr_viewpoints - sample_locations_count)]
+            sample_idx_select = torch.randperm(sample_idx.sum())[
+                : min(nbr_samples_per_point, nbr_viewpoints - sample_locations_count)
+            ]
             sample_locations[
                 sample_locations_count : sample_locations_count + sample_idx_select.shape[0]
             ] = self.terrain_analyser.samples[sample_idx][sample_idx_select, :2]
@@ -165,7 +167,11 @@ def render_viewpoints(self, samples: torch.Tensor):
         np.savetxt(os.path.join(filedir, "intrinsics.txt"), intrinsics.reshape(-1, 12), delimiter=",")
 
         # save camera poses (format: x y z qx qy qz qw instead of x y z qw qx qy qz)
-        np.savetxt(os.path.join(filedir, "camera_extrinsic.txt"), samples[:, [0, 1, 2, 4, 5, 6, 3]].cpu().numpy(), delimiter=",")
+        np.savetxt(
+            os.path.join(filedir, "camera_extrinsic.txt"),
+            samples[:, [0, 1, 2, 4, 5, 6, 3]].cpu().numpy(),
+            delimiter=",",
+        )
 
         # save images
         samples = samples.to(self.scene.device)
@@ -217,12 +223,14 @@ def render_viewpoints(self, samples: torch.Tensor):
                             #       might not start from 0 as well as some data ids might not be present in the label ids
                             unique_data_ids = np.unique(image_data_np)
                             unique_data_ids.sort()
-                            mapping = np.zeros((max(unique_data_ids.max() + 1, max(info.keys()) + 1), 3), dtype=np.uint8)
+                            mapping = np.zeros(
+                                (max(unique_data_ids.max() + 1, max(info.keys()) + 1), 3), dtype=np.uint8
+                            )
                             mapping[list(info.keys())] = np.array(list(info.values()), dtype=np.uint8)
                             output = mapping[image_data_np[idx].squeeze(-1)]
                         else:
                             output = image_data_np[idx]
-                        
+
                         assert cv2.imwrite(
                             os.path.join(filedir, "semantics", f"{image_idx[cam_idx]}".zfill(4) + ".png"),
                             cv2.cvtColor(output.astype(np.uint8), cv2.COLOR_RGB2BGR),
@@ -233,8 +241,11 @@ def render_viewpoints(self, samples: torch.Tensor):
                             os.path.join(filedir, "depth", f"{image_idx[cam_idx]}".zfill(4) + ".png"),
                             np.uint16(image_data_np[idx] * self.cfg.depth_scale),
                         )
-                        # save as npy 
-                        np.save(os.path.join(filedir, "depth", f"{image_idx[cam_idx]}".zfill(4) + ".npy"), image_data_np[idx] * self.cfg.depth_scale)
+                        # save as npy
+                        np.save(
+                            os.path.join(filedir, "depth", f"{image_idx[cam_idx]}".zfill(4) + ".npy"),
+                            image_data_np[idx] * self.cfg.depth_scale,
+                        )
 
                     image_idx[cam_idx] += 1
 

omniverse/extension/omni.viplanner/omni/viplanner/utils/unreal_importer.py

@@ -151,7 +151,9 @@ def _add_semantics(self):
         # assign class to mesh in ISAAC
         def recursive_semUpdate(prim, sem_class_name: str, update_submesh: bool) -> bool:
             # Necessary for Park Mesh
-            if prim.GetName() == "HierarchicalInstancedStaticMesh" or prim.GetTypeName() == "Mesh":  # or "FoliageInstancedStaticMeshComponent" in prim.GetName():
+            if (
+                prim.GetName() == "HierarchicalInstancedStaticMesh" or prim.GetTypeName() == "Mesh"
+            ):  # or "FoliageInstancedStaticMeshComponent" in prim.GetName():
                 add_update_semantics(prim, sem_class_name)
                 update_submesh = True
             children = prim.GetChildren()

omniverse/standalone/data_collect.py

@@ -73,7 +73,7 @@ def main():
         cfg.terrain_analysis.dim_limiter_prim = "Road_Sidewalk"
     elif args_cli.scene == "warehouse":
         scene_cfg = WarehouseTerrainSceneCfg(args_cli.num_envs, env_spacing=1.0)
-        # overwrite semantic cost mapping 
+        # overwrite semantic cost mapping
         cfg.terrain_analysis.semantic_cost_mapping = CarlaSemanticCostMapping()
         # limit space to be within the road network
         cfg.terrain_analysis.dim_limiter_prim = "Section"  # name of the meshes of the walls

viplanner/config/costmap_cfg.py

@@ -158,13 +158,13 @@ class GeneralCostMapConfig:
     # smoothing parameters
     sigma_smooth: float = 3.0
     # cost map expansion
-    x_min: Optional[float] = None  # -8.05
+    x_min: Optional[float] = None
     # [m] if None, the minimum of the point cloud is used None (carla town01:  -8.05   matterport: None)
-    y_min: Optional[float] = None  # -8.05
+    y_min: Optional[float] = None
     # [m] if None, the minimum of the point cloud is used None (carla town01:  -8.05   matterport: None)
-    x_max: Optional[float] = None  #  346.22
+    x_max: Optional[float] = None
     # [m] if None, the maximum of the point cloud is used None (carla town01:  346.22  matterport: None)
-    y_max: Optional[float] = None  # 336.65
+    y_max: Optional[float] = None
     # [m] if None, the maximum of the point cloud is used None (carla town01:  336.65  matterport: None)
 
 

viplanner/depth_reconstruct.py

@@ -33,7 +33,7 @@ class DepthReconstruction:
         - semantics (optional)
             - xxxx.png  (images should be named with 4 digits, e.g. 0000.png, 0001.png, etc., RGB images)
 
-    In the case that the semantic and depth images have an offset in their position (as typical on some robotic platforms), 
+    In the case that the semantic and depth images have an offset in their position (as typical on some robotic platforms),
     define a sem_suffic and depth_suffix in ReconstructionCfg to differentiate between the two with the following structure:
 
     - env_name