unit test added

opencda/core/sensing/perception/sensor_transformation.py

@@ -109,14 +109,17 @@ def bbx_to_world(cords, vehicle):
  """
  Convert bounding box coordinate at vehicle reference to world reference.
  Args:
- cords (np.ndarray): Bounding box coordinates with 8 vertices.
+ cords (np.ndarray): Bounding box coordinates with 8 vertices, shape (n, 4)
  vehicle (carla.vehicle or ObstacleVehicle): vehicle object.
 
  Returns:
  bb_world_cords: np.ndarray
  Bounding box coordinates under word reference.
  """
- bb_transform = carla.Transform(vehicle.bounding_box.location)
+ if hasattr(vehicle.bounding_box, 'transform'):
+ bb_transform = vehicle.bounding_box.transform
+ else:
+ bb_transform = carla.Transform(vehicle.bounding_box.location)
  # bounding box to vehicle transformation matrix
  bb_vehicle_matrix = x_to_world_transformation(bb_transform)
 
@@ -169,31 +172,31 @@ def vehicle_to_sensor(cords, vehicle, sensor_transform):
  """
  Transform coordinates from vehicle reference to sensor reference
  Args:
- cords (np.ndarray): Coordinates under vehicle reference, shape (4, n)
+ cords (np.ndarray): Coordinates under vehicle reference, shape (n, 4)
  vehicle (carla.vehicle or ObstacleVehicle): vehicle object.
  sensor_transform (carla.Transform): sensor position in the world, shape(3, 1)
 
  Returns:
- (np.ndarray): Coordinates in sensor reference.
+ (np.ndarray): Coordinates in sensor reference, shape(4, n).
  """
  world_cord = bbx_to_world(cords, vehicle)
  sensor_cord = world_to_sensor(world_cord, sensor_transform)
 
  return sensor_cord
 
 
-def get_bounding_box(vehicle, sensor, sensor_transform):
+def get_bounding_box(vehicle, camera, sensor_transform):
  """
  Get vehicle bounding box and project to sensor image
  Args:
  vehicle (carla.vehicle or ObstacleVehicle): vehicle object.
- sensor (carla.sensor.camera.rgb): The CARLA sensor object.
+ camera (carla.sensor.camera.rgb): The CARLA sensor object.
  sensor_transform (carla.Transform): sensor position in the world
 
  Returns:
  (np.ndarray): Bounding box coordinates in sensor image.
  """
- camera_k_matrix = get_camera_intrinsic(sensor)
+ camera_k_matrix = get_camera_intrinsic(camera)
  # bb_cords is relative to bbx center(approximate the vehicle center)
  bb_cords = create_bb_points(vehicle)
 

test/mocked_carla.py

@@ -0,0 +1,102 @@
+# -*- coding: utf-8 -*-
+"""
+Mock Carla for unit tests.
+"""
+
+# Author: Runsheng Xu <[email protected]>
+# License: MIT
+
+import numpy as np
+
+
+class Location(object):
+ """ A mock class for Location. """
+
+ def __init__(self, x, y, z):
+ self.x = x
+ self.y = y
+ self.z = z
+
+
+class Transform(object):
+ """A mock class for transform"""
+
+ def __init__(self, x, y, z, pitch=0, yaw=0, roll=0):
+ self.location = Location(x, y, z)
+ self.rotation = Rotation(pitch, yaw, roll)
+
+
+class Rotation(object):
+ """ A mock class for Rotation. """
+
+ def __init__(self, pitch, yaw, roll):
+ self.pitch = pitch
+ self.yaw = yaw
+ self.roll = roll
+
+
+class Vector3D(object):
+ """ A mock class for Vector3D. """
+
+ def __init__(self, x, y, z):
+ self.x = x
+ self.y = y
+ self.z = z
+
+
+class Camera(object):
+ """A mock class for camera. """
+
+ def __init__(self, attributes: dict):
+ self.attributes = attributes
+ self.transform = Transform(x=10, y=10, z=10)
+
+ def get_transform(self):
+ return self.transform
+
+
+class Lidar(object):
+ """A mock class for lidar."""
+
+ def __init__(self, attributes: dict):
+ self.attributes = attributes
+ self.transform = Transform(x=11, y=11, z=11)
+
+ def get_transform(self):
+ return self.transform
+
+
+class BoundingBox(object):
+ """
+ A mock class for bounding box.
+ """
+
+ def __init__(self, corners):
+ """
+ Construct class.
+ Args:
+ corners (nd.nparray): Eight corners of the bounding box. shape:(8, 3)
+ """
+ center_x = np.mean(corners[:, 0])
+ center_y = np.mean(corners[:, 1])
+ center_z = np.mean(corners[:, 2])
+
+ extent_x = (np.max(corners[:, 0]) - np.min(corners[:, 0])) / 2
+ extent_y = (np.max(corners[:, 1]) - np.min(corners[:, 1])) / 2
+ extent_z = (np.max(corners[:, 2]) - np.min(corners[:, 2])) / 2
+
+ self.location = Location(x=center_x, y=center_y, z=center_z)
+ self.transform = Transform(x=center_x, y=center_y, z=center_z)
+ self.extent = Vector3D(x=extent_x, y=extent_y, z=extent_z)
+
+
+class Vehicle(object):
+ """A mock class for vehicle"""
+
+ def __init__(self):
+ corner = np.random.random((8, 3))
+ self.transform = Transform(x=12, y=12, z=12)
+ self.bounding_box = BoundingBox(corner)
+
+ def get_transform(self):
+ return self.transform

test/test_sensor_transformation.py

@@ -0,0 +1,76 @@
+# -*- coding: utf-8 -*-
+"""
+Unit test for sensor transformation
+"""
+# Author: Runsheng Xu <[email protected]>
+# License: MIT
+
+import os
+import sys
+import unittest
+
+# temporary solution for relative imports in case opencda is not installed
+# if opencda is installed, no need to use the following line
+sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), '../opencda')))
+sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), '.')))
+
+import mocked_carla as mcarla
+from opencda.core.sensing.perception.sensor_transformation import *
+
+
+class TestSensorTransformation(unittest.TestCase):
+ def setUp(self):
+ # random cords, [(x, y, z, 1), n]
+ self.cords = np.random.random(size=(4, 10))
+ self.vehicle = mcarla.Vehicle()
+ self.camera = mcarla.Camera({'image_size_x': 600,
+ 'image_size_y': 800,
+ 'fov': 90})
+ self.lidar = mcarla.Lidar({'channels': 32,
+ 'range': 50})
+
+ self.camera_transform = mcarla.Transform(x=11, y=11, z=11)
+ self.lidar_transform = mcarla.Transform(x=10, y=10, z=10)
+
+ self.rgb_image = np.random.randint(0, 255, size=(800, 600, 3)).astype('uint8')
+ self.point_cloud = np.random.random(size=(100, 4))
+
+ def test_x_to_world_transformation(self):
+ assert x_to_world_transformation(self.lidar_transform).shape == (4, 4)
+ assert x_to_world_transformation(self.lidar_transform)[3, 3] == 1
+
+ def test_world_to_sensor(self):
+ assert world_to_sensor(self.cords, self.lidar_transform).shape == (4, self.cords.shape[1])
+
+ def test_sensor_to_world(self):
+ assert sensor_to_world(self.cords, self.lidar_transform).shape == (4, self.cords.shape[1])
+
+ def test_get_camera_intrinsic(self):
+ assert get_camera_intrinsic(self.camera).shape == (3, 3)
+ assert get_camera_intrinsic(self.camera)[2, 2] == 1
+
+ def test_create_bb_points(self):
+ assert create_bb_points(self.vehicle).shape == (8, 4)
+ assert create_bb_points(self.vehicle)[:, 3].all() == 1
+
+ def test_bbx_to_world(self):
+ assert bbx_to_world(self.cords.T, self.vehicle).shape == (4, self.cords.shape[1])
+
+ def test_vehicle_to_sensor(self):
+ assert vehicle_to_sensor(self.cords.T, self.vehicle, self.camera_transform).shape == (4, self.cords.shape[1])
+
+ def test_get_bounding_box(self):
+ assert get_bounding_box(self.vehicle, self.camera, self.camera_transform).shape == (8, 3)
+
+ def test_get_2d_bb(self):
+ assert get_2d_bb(self.vehicle, self.camera, self.camera_transform).shape == (2, 2)
+
+ def test_project_lidar_to_camera(self):
+ assert project_lidar_to_camera(self.lidar, self.camera, self.point_cloud, self.rgb_image)[1].shape == \
+ (self.point_cloud.shape[0], 3)
+ assert project_lidar_to_camera(self.lidar, self.camera, self.point_cloud, self.rgb_image)[0].shape == \
+ self.rgb_image.shape
+
+
+if __name__ == '__main__':
+ unittest.main()