Skip to content

Latest commit

 

History

History
125 lines (85 loc) · 4.46 KB

error_resolution_en.md

File metadata and controls

125 lines (85 loc) · 4.46 KB
Raw

Error resolution

Errors at "setup_ubuntu20.04.sh"

Fail at TASK [cuda : CUDA (add CUDA repository into sources.list)]

image

The following command resolved this issue.

sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/3bf863cc.pub

Fail at TASK [cuda : install software-properties-common]

Screenshot from 2022-05-16 10-49-50

The cause is a key update in the NVIDIA repository. The following command resolved the issue.

sudo apt-key del 7fa2af80
wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/cuda-keyring_1.0-1_all.deb
sudo dpkg -i cuda-keyring_1.0-1_all.deb 
sudo apt update

※If the site's procedures and error solutions do not work, ，Add XXXXXXX public key

sudo apt-key adv --keyserver keyserver.ubuntu.com --recv-keys XXXXXX

Fail at TASK [tensorrt : TensorRT (install libraries for TensorRT 7)]

Screenshot from 2022-05-23 23-12-47

The cause is that the version of CUDA is different from the required version. In my case, the following procedure solved the problem.

※cuda: 11.1, cudnn: 8.0.5, TensorRT: 7.2.1

sudo apt remove nvidia-*
sudo dpkg --remove --force-remove-reinstreq XXXXXXX　※Forcibly remove the not upgraded package in the lower part of the error message in the attached image.

Fail at TASK [libtorch : Libtorch (unarchive)]

Screenshot from 2022-05-23 23-25-46

The following procedure resolves the problem.

  1. Download this package

  2. Rename the downloaded zip file to libtorch.zip.

  3. Move libtorch.zip to /tmp

  4. Comment out 2~5 lines in　/aichallenge2021final/ansible/roles/libtorch/tasks/main.yaml

Errors at "colcon build"

Failed <<< tier4_planning_rviz_plugin

Screenshot from 2022-05-24 08-53-40

Change the following in　/aichallenge2021final/src/autoware/universe/common/tier4_planning_rviz_plugin/src/tools/jsk_overlay_utils.hpp

#include <QImage>
        ↓Change to the following
#include <QCursor>
#include <QImage>
#include <QVariant>

Errors at "PSim / LSim"

Error at [rmw_cyclonedds_cpp]

 Screenshot from 2022-05-26 22-40-13

  1. Change the DDS to be used
export RMW_IMPLEMENTATION=rmw_cyclonedds_cpp
sudo apt install ros-foxy-rmw-cyclonedds-cpp
  1. Setting up cyclonedds to do multicast on localhost

  The file can be placed anywhere, but let's place it in /opt/autoware/cyclonedds_config.xml and copy & paste the following contents

 <?xml version="1.0" encoding="UTF-8" ?>
 <CycloneDDS xmlns="https://cdds.io/config" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="https://cdds.io/config https://raw.githubusercontent.com/eclipse-cyclonedds/cyclonedds/master/etc/cyclonedds.xsd">
   <Domain id="any">
     <General>
       <NetworkInterfaceAddress>lo</NetworkInterfaceAddress>
     </General>
     <Internal>
       <MinimumSocketReceiveBufferSize>10MB</MinimumSocketReceiveBufferSize>
     </Internal>
   </Domain>
 </CycloneDDS>
  1. After creating the file, add the following to ~/.bashrc
 export CYCLONEDDS_URI=file:///opt/autoware/cyclonedds_config.xml
  1. Change maximum receive buffer size (required on every restart)
sudo sysctl -w net.core.rmem_max=2147483647
  1. Multicast support for localhost (required on every reboot)
sudo ifconfig lo multicast