Using Problem: After runing the package (docker version ), checking ros topic, sensor topic frame is not compatible with tf tree

[LifeIsBinary]

Thanks for this work.

I get some problem following the tutorial in readme. Any help will be appreicated.

1. After starting carla0.9.15(localhost) and the package(docker), starting autoware(localhost), the rviz2 panel open and close frequently. I catch the message " cannot receive pointcloud" in the output log when starting autoware.

2. So i intent to view the pointcloud and image topic, only Qos setting Best Effort and fixed frame lidar_front can make two topic data visualable.

I realized that the sensor data and tf topic published is dismatch in frame. sensor frame (as expect):

• camera_front_link
• lidar_front

• lidar_rear

  tf topic(may be abnormal):

I want know the following things: a. First, the dismatch is normal? If not , is there a way to fix the dismatch? b. Is it right that autoware can not work as expect due to the frame dismatch?

my enviroment:

• memory :64G GPU:nvidia RTX3090 processor:inter i9-11900K
• ubuntu22.04
• ros-hunble
• cuda--11.7
• cudnn-8.9.5.30
• tensort-8.6.0.12
• autoware tag: v1.0

Finally I attach the output in terminal when starting autoware. autoware_carla.log

[Robotics010]

Hey @LifeIsBinary

Could you post your launching commands, so that I will be able to reproduce the same on my end?

Are you sure, that you have the map properly loaded?

[LifeIsBinary]

Hey @LifeIsBinary

Could you post your launching commands, so that I will be able to reproduce the same on my end?

Are you sure, that you have the map properly loaded?

Thanks for your reply so instantly ~

My launching commands as follows:

1. starting carla, carla -carla-rpc-port=1403 -prefernvidia
2. starting the repository package, sudo docker run -it -e RMW_IMPLEMENTATION=rmw_cyclonedds_cpp --network host tumgeka/carla-autoware-bridge:latest and ros2 launch carla_autoware_bridge carla_aw_bridge.launch.py port:=1403 town:=Town10HD
3. starting autoware, ros2 launch autoware_launch e2e_simulator.launch.xml vehicle_model:=carla_t2_vehicle sensor_model:=carla_t2_sensor_kit map_path:=/home/lifeisbinary/autoware_map/carla_map/Town10

   Town10HD map was downloaded by readme link saving in /home/lifeisbinary/autoware_map/carla_map/Town10 directory

the diretory files list:

I am not sure but i can see map in rivz , like this: Because of uping and offing of riviz2 requently, I can not interact with it to check detail of map direcotry in rviz panel , to make sure that the status of osm and pcd map is OK. It seems that no point cloud layer, but i am not sure the layer visibility is close default or other reasons.

Could you have other way to check map loading sucessfully without rviz panel?

[Robotics010]

1. starting carla0.9.15(localhost) and the package(docker), starting autoware(localhost)

Could you try starting from docker (not host)?

Here is how I launch it:

robo@computer:~$ docker run --privileged --gpus all --net=host -e DISPLAY=$DISPLAY carlasim/carla:0.9.15 /bin/bash ./CarlaUE4.sh -carla-rpc-port=1403 -quality-level=Low

robo@computer:~$ docker run -it -e RMW_IMPLEMENTATION=rmw_cyclonedds_cpp --network host tumgeka/carla-autoware-bridge:latest
root@computer:/tum# ros2 launch carla_autoware_bridge carla_aw_bridge.launch.py port:=1403 town:=Town10HD timeout:=10

robo@computer:~$ rocker --network=host -e RMW_IMPLEMENTATION=rmw_cyclonedds_cpp -e LIBGL_ALWAYS_SOFTWARE=1 --x11 --nvidia --volume /home/robo/autoware -- ghcr.io/autowarefoundation/autoware-universe:humble-2024.01-cuda-amd64
root@computer:/home/robo/autoware/autoware# ros2 launch autoware_launch e2e_simulator.launch.xml vehicle_model:=carla_t2_vehicle sensor_model:=carla_t2_sensor_kit map_path:=/home/robo/autoware/Town10

[LifeIsBinary]

1. starting carla0.9.15(localhost) and the package(docker), starting autoware(localhost)

Could you try starting from docker (not host)?

Hi~, guys. Today I download carla0.9.15((docker) and autoware(docker), building the package to docker image, and rebuilding autoware code in autoware docker following repository readme.

Starting all from docker, I face the same situation.

The rviz2 panel still up and down frequently between figure1 and figure2.

The terminal output of autoware docker is "[perception.object_recognition.detection.obstacle_pointcloud_based_validator_node]: cannot receive pointcloud". I found the message in bridge terminal, "[bridge-1] [WARN] [1720019146.814677533] [carla_ros_bridge]: New subscription discovered on topic '/carla/ego_vehicle/sensor/lidar/front', requesting incompatible QoS. No messages will be sent to it. Last incompatible policy: RELIABILITY". I think that it makes step3 abnormal. But how to fix it ?

adding timeout parameter to Step2 launch command(starting bridge in docker), make "cannot reveive pointcloud " and "incompatible Qos" go out. But getting new Error in Step3(autoware), "vehicle_state_report msg is timeout"

The log file is here(use timeout=10 in step2): autoware_launch_terminal_ouput.log

PS: I check the tf topic, after autoware starting , frame in tf and data (lidar,camera) match. I see that the matching tf is published in Step3.