<Frame> passed to lookupTransform argument t arget_frame does not exist

[songanz]

Bug report

Required Info:

• Operating System:
  • ROS1 melodic and Gazebo running in docker with Ubuntu 18.04
  • ROS2 foxy running in docker with Ubuntu 20.04
• Installation type:
  • Build from source
• Version or commit hash:
  • version 0.10.2
• DDS implementation:
  • No
• Client library (if applicable):
  • rclpy

Steps to reproduce issue

1. Start any gazebo simulator you have which contains at least two frames on a ROS1 docker
   1. Set the use_sim_time to be "true" in that simulator
   2. I am particularly using this environment from CMU: cmu_env with roslaunch vehicle_simulator system_campus.launch
2. Start the ros1_bridge in a ROS2 docker

export ROS_MASTER_URI=http://localhost:11311
ros2 run ros1_bridge dynamic_bridge --bridge-all-topics --ros-args -p use_sim_time:=true

3. Set up tf_listener in ROS2

class TFListener(Node):
    def __init__(self) -> None:
        super().__init__('cam_sem_seg')
        """ parameters """
        self.declare_parameters(
            namespace='cam_sem_seg_py',
            parameters=[
                ('use_sim_time', None),
                ('image_topic', None),
                ('camera_info', None),
                ('lidar_topic', None),
                ('camera_frame', None),
                ('lidar_frame', None),
            ]
        )
        use_sim_time = self.get_parameter('cam_sem_seg_py.use_sim_time')
        # set use_sim_time to be true in your launch
        use_sim_time = rclpy.parameter.Parameter('use_sim_time', rclpy.Parameter.Type.BOOL, use_sim_time.value)  
        self.set_parameters([use_sim_time])

        """ ros """        
        """ sub/pub """
        (image_topic, camera_info, lidar_topic) = self.get_parameters(
            ['cam_sem_seg_py.image_topic', 'cam_sem_seg_py.camera_info', 'cam_sem_seg_py.lidar_topic'])
        
        self.img_sub = message_filters.Subscriber(self, Image, image_topic.value)
        self.info_sub = message_filters.Subscriber(self, CameraInfo, camera_info.value)
        self.lidar_sub = message_filters.Subscriber(self, PointCloud2, lidar_topic.value)
        
        ts = message_filters.ApproximateTimeSynchronizer([self.img_sub, self.info_sub, self.lidar_sub], 10, 0.04)
        ts.registerCallback(self.call_back)

        # TF listener
        self.tf_buffer = Buffer()
        self.tf_listener = TransformListener(self.tf_buffer, self)
        (self.camera_frame, self.lidar_frame) = self.get_parameters(
            ['cam_sem_seg_py.camera_frame', 'cam_sem_seg_py.lidar_frame'])

        self.pub = self.create_publisher(PointCloud2, "/lidar_label", 10)
        

    def call_back(self, img_msg, info_sub, lidar_msg):
        # load intrinsic
        self.intrinsic = np.array(info_sub.p).reshape(3,4)  # if error, double check the camera_info
        
        # get extrinsic
        try:
            lid2cam_T = self.tf_buffer.lookup_transform(
                self.camera_frame.value,  # To
                self.lidar_frame.value,  # from
                self.get_clock().now())
            self.extrinsic = np.dot(lid2cam_T, self.intrinsic)
            
        except TransformException as ex:
            print('use_sim_time: ', self.get_parameters(['use_sim_time'])[0].value)
            print('self.get_clock().now(): ', self.get_clock().now())
            print('rclpy.time.Time(): ', rclpy.time.Time())
            self.get_logger().info(
                f'Could not transform {self.camera_frame.value} to {self.lidar_frame.value}: {ex}')
            return
        """ Process the image and lidar messages """
        ......
        self.pub.publish(pc2)

4. Launch file for the tf_listener

from launch import LaunchDescription
from launch_ros.actions import Node
from launch.substitutions import LaunchConfiguration
from ament_index_python.packages import get_package_share_directory

def generate_launch_description():
    use_sim_time = LaunchConfiguration('use_sim_time', default='true')
    package_name='your_name'
    your_node = Node(
        package= package_name,
        namespace= 'your_namespace',
        executable= 'your_executable',
        name= package_name,
        parameters=[
            {'use_sim_time': use_sim_time},
            {'image_topic': '/camera/image'},
            {'camera_info': '/camera/camera_info'},
            {'lidar_topic': '/sensor_scan'},
            {'camera_frame':'camera'},
            {'lidar_frame': 'sensor_at_scan'},
        ],
        emulate_tty=True
    )
    return LaunchDescription([your_node])

5. On ROS2 docker terminal: ros2 run tf2_ros tf2_echo <reference_frame> <target_frame>
   You will see it cannot find the frame first, and then it finds the frame with a completely different timestamp.
   
6. In the meantime, the tf_listener is showing:
   
   which means even though I use self.get_clock.now() (it is the sim_time) in the lookup_transform() function, it still cannot find the frame.

Expected behavior

Find the frame and get the TF correctly

Actual behavior

As described in the "Steps to reproduce issue" part

Additional information

NAN

[songanz]

Any comment on this issue?