Autoware 安装（踩坑指南）
【Autoware】2小时安装Autoware1.13（保姆级教程）
Autoware入门学习（二）——Ubuntu18.04下的源码安装和配置
上面的两篇博客安装都异常顺利，甚至没有一点报错，实际安装中显然是不可能的
安装环境
- Ubuntu 18.04
- ROS Melodic
- Qt 5.10.0
- OpenCV 3.4.16
看一下官网的 Requirements
这里 CUDA 是选装的，CUDA 要求是10.0的，而我是11的，而且 Eigen 版本也不一样，所以编译时没有选择 CUDA，后面有实际需要再说
下载 autoware.ai.repos 时出错，怎么都下载不下来
wget -O autoware.ai.repos "https://raw.githubusercontent.com/autowarefoundation/autoware_ai/1.12.0/autoware.ai.repos"
直接去查看这个文件，复制到 autoware.ai.repos 即可
repositories:
autoware/common:
type: git
url: https://gitlab.com/autowarefoundation/autoware.ai/common.git
version: 1.12.0
autoware/core_perception:
type: git
url: https://gitlab.com/autowarefoundation/autoware.ai/core_perception.git
version: 1.12.0
autoware/core_planning:
type: git
url: https://gitlab.com/autowarefoundation/autoware.ai/core_planning.git
version: 1.12.0
autoware/documentation:
type: git
url: https://gitlab.com/autowarefoundation/autoware.ai/documentation.git
version: 1.12.0
autoware/messages:
type: git
url: https://gitlab.com/autowarefoundation/autoware.ai/messages.git
version: 1.12.0
autoware/simulation:
type: git
url: https://gitlab.com/autowarefoundation/autoware.ai/simulation.git
version: 1.12.0
autoware/utilities:
type: git
url: https://gitlab.com/autowarefoundation/autoware.ai/utilities.git
version: 1.12.0
autoware/visualization:
type: git
url: https://gitlab.com/autowarefoundation/autoware.ai/visualization.git
version: 1.12.0
drivers/awf_drivers:
type: git
url: https://gitlab.com/autowarefoundation/autoware.ai/drivers.git
version: 1.12.0
citysim:
type: git
url: https://github.com/CPFL/osrf_citysim.git
version: 27bd05bc6c762b3ad8c9bb85f678d4b7ce7a27c5
car_demo:
type: git
url: https://github.com/CPFL/car_demo.git
version: e364448fad421cb6244c9f828f978d8d877dcbf9
drivers/ds4:
type: git
url: https://github.com/tier4/ds4.git
version: 5aa2f7f53a67992fffa7c801ed9c663a380b5d4a
rosdep install 的时候提示缺少相关包
rosdep install -y --from-paths src --ignore-src --rosdistro $ROS_DISTRO
缺什么安装什么，例如 sudo apt-get install ros-melodic-nmea-msgs
redwall@redwall-G3-3500:~/autoware$ rosdep install -y --from-paths src --ignore-src --rosdistro $ROS-DISTRO
WARNING: given --rosdistro -DISTRO but ROS_DISTRO is "melodic". Ignoring environment.
WARNING: ROS_PYTHON_VERSION is unset. Defaulting to 2
ERROR: the following packages/stacks could not have their rosdep keys resolved
to system dependencies:
autoware_bag_tools: Cannot locate rosdep definition for [nmea_msgs]
kitti_box_publisher: Cannot locate rosdep definition for [jsk_recognition_msgs]
range_vision_fusion: Cannot locate rosdep definition for [jsk_topic_tools]
runtime_manager: Cannot locate rosdep definition for [nmea_navsat_driver]
as: Cannot locate rosdep definition for [pacmod_msgs]
autoware_camera_lidar_calibrator: Cannot locate rosdep definition for [image_view2]
lidar_euclidean_cluster_detect: Cannot locate rosdep definition for [jsk_rviz_plugins]
没用 CUDA 用下面的语句编译
colcon build --cmake-args -DCMAKE_BUILD_TYPE=Release
这里遇到了很多问题，可以归类两个大问题
多版本 OpenCV 问题
安装 ROS 时会安装 3.2.0 版本的 OpenCV，后面自己又安装了 3.4.16 版本的 OpenCV
之前不晓得从哪里看来的博客，两个版本都安装到了 /usr/local/ 目录下，那么后安装的 3.4.16 版本的 OpenCV 就会覆盖之前的 3.2.0 版本的
那么就会导致版本的混乱，同时在 build 目录下 sudo make uninstall 时会把 3.2.0 版本的也一起卸载掉，这就导致很多 ROS 基础的功能包都找不到 OpenCV 头文件
ubuntu卸载opencv,简单快速，亲测有效
ubuntu18.04中多版本OpenCV切换的尝试
ubuntu自定义路径安装OpenCV
卸载重装 OpenCV 3.4.16，注意制定安装路径为 /usr/local/opencv-3.4.16
cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local/opencv-3.4.16 ..
make 编译前可以 nproc 查看处理器核心数量，充分利用核心数量进行多线程编译
安装完成后在bashrc文件末尾增加如下语句
export PKG_CONFIG_PATH=/usr/local/opencv-3.4.16/lib/pkgconfig
export LD_LIBRARY_PATH=/usr/local/opencv-3.4.16/lib
export OpenCV_DIR=/usr/local/opencv-3.4.16
加入前两条语句后，pkg-config能发现opencv
加入第三条语句后，能在CMakeLists.txt中直接通过FIND_PACKAGE(OpenCV REQUIRED)命令找到opencv
对于 OpenCV 3.2.0，也是重新安装，用下述命令卸载
sudo apt-get remove 'libopencv*'
会随着卸载很多软件包，记录下来，再重新安装，OpenCV 3.2.0 用下述命令重新安装
2023-06-16.txt
sudo apt-get install 'libopencv*'
至此，编译时报错对 cv::Mat() 未定义的引用解决
OpenCV版本不兼容
代码中opencv2风格的代码与opencv3风格代码混用，报了很多错误
/home/fengbro/autoware.ai/src/autoware/core_perception/vision_lane_detect/nodes/vision_lane_detect/vision_lane_detect.cpp:514:58: error: could not convert ‘cv::Scalar_((double)0, (double)255, (double)255, (double)0)’ from ‘cv::Scalar {aka cv::Scalar_}’ to ‘CvScalar’
cvPoint(frame_size.width/2, frame_size.height), CV_RGB(255, 255, 0), 1);
vision_lane_detect.cpp文件，修改其中第514行代码：
CV_RGB(255, 255, 0) —> cvScalar(255, 255, 0)
/home/fengbro/autoware.ai/src/autoware/core_perception/vision_lane_detect/nodes/vision_lane_detect/vision_lane_detect.cpp:544:30: error: conversion from ‘cv::Mat’ to non-scalar type ‘IplImage {aka _IplImage}’ requested
IplImage frame = cv_image->image;
vision_lane_detect.cpp文件的第544行中：
IplImage frame = cv_image->image;-----> IplImage frame = cvIplImage(cv_image->image);
/home/fengbro/autoware.ai/src/autoware/core_perception/vision_darknet_detect/src/vision_darknet_detect.cpp:228:17: error: no match for ‘operator=’ (operand types are ‘IplImage {aka _IplImage}’ and ‘cv::Mat’)
ipl_image = final_mat;
修改vision_darknet_detect.cpp文件的第228行：
ipl_image = final_mat; 为 ipl_image = cvIplImage(final_mat);
/home/fengbro/autoware.ai/src/autoware/core_perception/vision_beyond_track/include/detection.h:234:21: error: conversion from ‘cv::Mat’ to non-scalar type ‘CvMat’ requested
CvMat cvmat = sum_mat;
在detection.h文件的第234行下一行添加如下内容
for(size_t i=0; i< sum_mat.rows; ++i)
for(size_t j=0; j< sum_mat.cols; ++j)
((double*)(cvmat.data.ptr + i*cvmat.step))[j] = sum_mat.at<double>(i,j);
/home/fengbro/autoware.ai/src/autoware/core_perception/trafficlight_recognizer/nodes/region_tlr/region_tlr.cpp:143:63: error: no match for ‘operator=’ (operand types are ‘CvPoint’ and ‘cv::Point {aka cv::Point_<int>}’)
textOrg = cv::Point(ctx.topLeft.x, ctx.botRight.y + baseline);
修改region_tlr.cpp文件的第143行如下文章来源：https://www.toymoban.com/news/detail-492051.html
textOrg = cvPoint(ctx.topLeft.x, ctx.botRight.y + baseline);
顺利安装后会提示如下信息文章来源地址https://www.toymoban.com/news/detail-492051.html
---
Finished <<< trafficlight_recognizer [18.1s]
Summary: 139 packages finished [52.9s]
到了这里，关于Autoware 安装（踩坑指南）的文章就介绍完了。如果您还想了解更多内容，请在右上角搜索TOY模板网以前的文章或继续浏览下面的相关文章，希望大家以后多多支持TOY模板网！
