Issue about unsuccessful 2D pose estimate when work with autoware

[hangger-258]

Although I install the AWSIM and Autoware follow the Quick start demo, In the STEP5. Let's Self-Driving simulation, Launch both the AWSIM and Autoware.
I can't Automaticalely set 2D Pose Estimate with RViz as the tutorial. The state panel show that it is always in INITIALIZING. And when I set the 2D goal post, there is no path to be generated.
Additionally, The autoware can work itself in the official tutorial of autoware.

[shmpwk]

@hangger-258
Could you share your demo video and log files of the console output under ~/.ros/log/[DATE, etc] (e.x. ~/.ros/log/2022-10-27-10-09-39-837180-npc2201036-2629664/).

[hangger-258]

https://drive.google.com/drive/folders/122-7FxNAzqCZ2_mYjpIJAekiXDMSaRdD?usp=sharing
Here is the link for videos and log files.

[shmpwk]

@hangger-258
Thank you.
Looking at your log file, there are several suspicious ERROR log.

For example, L24 says

1666844014.6560595 [ERROR] [map_hash_generator-9]: process has died [pid 19716, exit code 1, cmd '/home/human/Desktop/AWSIM-autoware/autoware/install/map_loader/lib/map_loader/map_hash_generator --ros-args -r __node:=map_hash_generator -r __ns:=/map --params-file /tmp/launch_params_7z7p_cgj --params-file /tmp/launch_params_scsb65ac'].

It corresponds to L138-L155.

1666844014.7487721 [map_hash_generator-9] Traceback (most recent call last):
1666844014.7557201 [map_hash_generator-9]   File "/home/human/Desktop/AWSIM-autoware/autoware/install/map_loader/lib/map_loader/map_hash_generator", line 21, in <module>
1666844014.7580757 [map_hash_generator-9]     from rclpy.node import Node
1666844014.7582481 [map_hash_generator-9]   File "/opt/ros/galactic/lib/python3.8/site-packages/rclpy/node.py", line 43, in <module>
1666844014.7590582 [map_hash_generator-9]     from rclpy.client import Client
1666844014.7591376 [map_hash_generator-9]   File "/opt/ros/galactic/lib/python3.8/site-packages/rclpy/client.py", line 22, in <module>
1666844014.7593884 [map_hash_generator-9]     from rclpy.impl.implementation_singleton import rclpy_implementation as _rclpy
1666844014.7594476 [map_hash_generator-9]   File "/opt/ros/galactic/lib/python3.8/site-packages/rclpy/impl/implementation_singleton.py", line 32, in <module>
1666844014.7596691 [map_hash_generator-9]     rclpy_implementation = import_c_library('._rclpy_pybind11', package)
1666844014.7604921 [map_hash_generator-9]   File "/opt/ros/galactic/lib/python3.8/site-packages/rpyutils/import_c_library.py", line 39, in import_c_library
1666844014.7633743 [INFO] [launch_ros.actions.load_composable_nodes]: Loaded node '/perception/object_recognition/detection/voxel_based_compare_map_filter' in container 'pointcloud_container'
1666844014.7642131 [map_hash_generator-9]     return importlib.import_module(name, package=package)
1666844014.7665758 [map_hash_generator-9]   File "/home/human/anaconda3/lib/python3.9/importlib/__init__.py", line 127, in import_module
1666844014.7667894 [map_hash_generator-9]     return _bootstrap._gcd_import(name[level:], package, level)
1666844014.7678573 [INFO] [launch_ros.actions.load_composable_nodes]: Loaded node '/default_ad_api/node/localization' in container '/default_ad_api/container'
1666844014.7699301 [map_hash_generator-9] ModuleNotFoundError: No module named 'rclpy._rclpy_pybind11'
1666844014.7721665 [INFO] [launch_ros.actions.load_composable_nodes]: Loaded node '/autoware_api/external/start' in container '/autoware_api/external/autoware_iv_adaptor'
1666844014.7732270 [map_hash_generator-9] The C extension '/opt/ros/galactic/lib/python3.8/site-packages/_rclpy_pybind11.cpython-39-x86_64-linux-gnu.so' isn't present on the system. Please refer to 'https://index.ros.org/doc/ros2/Troubleshooting/Installation-Troubleshooting/#import-failing-without-library-present-on-the-system' for possible solutions

It seems your python environment is mixed at ros galactic python3.8 and anaconda python3.9.
Could you check your python environment? python3.8 is recommended now.
Also, I heard that anaconda environment sometimes conflicts with ros python environment.

[hangger-258]

Thanks for your answer, actually the problem existed before I install anaconda, and I try to uninstall the anaconda and start the AWSIM-autoware again, the problem is still there.
Here is the log after uninstall the anaconda.
https://drive.google.com/file/d/1J9am9ZEglDA-HW9yJt2z_iUfP3pDwZv1/view?usp=sharing

[shmpwk]

@hangger-258
Thank you for your update.
Anaconda error disappears and it seems map_hash_generator correctly works.
However lidar_centerpoint node dies at L220

1666865292.2733142 [ERROR] [lidar_centerpoint_node-29]: process has died [pid 10292, exit code -6, cmd '/home/human/Desktop/AWSIM-autoware/autoware/install/lidar_centerpoint/lib/lidar_centerpoint/lidar_centerpoint_node --ros-args -r __node:=lidar_centerpoint -r __ns:=/perception/object_recognition/detection/centerpoint --params-file /tmp/launch_params_7fyt8v5w --params-file /tmp/launch_params_902az41u --params-file /tmp/launch_params_rya8jghl --params-file /tmp/launch_params_l8d4cp5o --params-file /tmp/launch_params_y76047qy --params-file /tmp/launch_params_vsbzfgia --params-file /tmp/launch_params_ql82ad0x --params-file /tmp/launch_params_1vifqbki --params-file /tmp/launch_params_02clc01e --params-file /tmp/launch_params_ps6y_pam --params-file /home/human/Desktop/AWSIM-autoware/autoware/install/lidar_centerpoint/share/lidar_centerpoint/config/centerpoint_tiny.param.yaml --params-file /home/human/Desktop/AWSIM-autoware/autoware/install/lidar_centerpoint/share/lidar_centerpoint/config/detection_class_remapper.param.yaml -r ~/input/pointcloud:=/sensing/lidar/concatenated/pointcloud -r ~/output/objects:=objects'].

This sometimes happens when your NVIDIA GPU environment is broken.
Could you share your enviroment info (ubuntu version, ros2 version, cpu, gpu)?
Also, could you share the output when executing nvidia-smi on your console?

[hangger-258]

Ubuntu version :5.15.0-52-generic #58~20.04.1-Ubuntu
Ros2: galactic
CPU 11th i9, GPU: 3070 8G
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 470.141.03 Driver Version: 470.141.03 CUDA Version: 11.4 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 NVIDIA GeForce ... Off | 00000000:01:00.0 On | N/A |
| 0% 32C P8 8W / 220W | 715MiB / 7979MiB | 1% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+

[shmpwk]

@hangger-258
Hmm... not sure but CUDA version is one possible cause of the problem.

Currently, we use CUDA11.6. https://github.com/autowarefoundation/autoware/blob/e108b8fa9e47307b1f2573be5d0dde828235bcb0/amd64.env

In the tutorial, if you execute ./setup-dev-env.sh without prepared CUDA environment, appropriate version of CUDA will be installed.
However, if you have prepared CUDA environment, it won't.

Your CUDA11.4 is not recommend but it might work...

[shmpwk]

@hangger-258
Could you try to add pcl_ros in universe/autoware.universe/sensing/pointcloud_preprocessor/CMakeLists.txt following https://github.com/autowarefoundation/autoware.universe/pull/2101/files and colcon build again?

[shmpwk]

↑This might be a solution
#15 (comment)

[hangger-258]

Thank you very much! I will try it.

[shmpwk]

I updated the awsim-stable branch by autowarefoundation/autoware#2982.

[shmpwk]

@hangger-258 How's it going?

[mackierx111]

This issue is closed. Please open it if necessary.