|
百度 Apollo 的简单编译运行方法
下载 apollo 源码并进入 apollo 目录
$ git clone https://github.com/ApolloAuto/apollo.git$ cd apollo$ sudo docker/setup_host/install_docker.sh
官方代码库中的 docker/setup_host/install_docker.sh 脚本由于没有可执行权限,上面的命令会执行失败。在执行上面的命令之前,还需要给这个脚本加上可执行权限:
$ chmod a+x docker/setup_host/install_docker.sh$ sudo docker/scripts/dev_start.sh$ sudo docker/scripts/dev_into.sh$ sudo ./apollo.sh build
这个命令编译的是 release 版的二进制文件,也就是说生成的二进制文件,包括动态链接库和可执行文件,不包含调试用的符号信息。把 apollo.sh 命令的 build 参数换成 build_dbg 参数,可以编译调试版二进制文件。
- 启动 Dreamview
在 docker 环境中执行如下命令:
# sudo scripts/bootstrap.sh
若正常启动,打开 http://localhost:8888 可以看到效果
Cyber RT 有一个示例应用演示了 Cyber RT 框架的基本用法,位于 apollo/cyber/examples/common_component_example。可以在 docker 环境中执行示例应用。
在执行下面的程序之前,可以先打开向标准错误输出的日志输出,具体来说可以将 cyber/setup.bash 脚本中的 GLOG_alsologtostderr 从 0 改为 1,并在当前终端中执行 source cyber/setup.bash。
或者也可以直接执行如下命令:
export GLOG_alsologtostderr=1
(1). 运行示例组件
可以通过如下命令执行示例组件:
# bazel-bin/cyber/tools/cyber_launch/cyber_launch start cyber/examples/common_component_example/common.launch
或者也可以直接通过 mainboard 来运行:
# bazel-bin/cyber/mainboard/mainboard -d cyber/examples/common_component_example/common.dag
此时可以在终端看到类似如下这样的输出:
root@in-dev-docker:/apollo# bazel-bin/cyber/mainboard/mainboard -d cyber/examples/common_component_example/common.dagWARNING: Logging before InitGoogleLogging() is written to STDERRI1216 19:24:43.487663 110641 module_argument.cc:81] []command: bazel-bin/cyber/mainboard/mainboard -d cyber/examples/common_component_example/common.dag I1216 19:24:43.488569 110641 global_data.cc:153] []host ip: 192.168.31.205W1216 19:24:43.488574 110641 environment.h:33] []Environment variable [CYBER_PATH] not set, fallback to I1216 19:24:43.489843 110641 module_argument.cc:57] []binary_name_ is mainboard, process_group_ is mainboard_default, has 1 dag confI1216 19:24:43.489851 110641 module_argument.cc:60] []dag_conf: cyber/examples/common_component_example/common.dagW1216 19:24:43.489900 110641 environment.h:33] [mainboard]Environment variable [CYBER_PATH] not set, fallback to W1216 19:24:43.489915 110641 scheduler_factory.cc:63] [mainboard]Scheduler conf named /apollo/cyber/conf/mainboard_default.conf not found, use default.W1216 19:24:43.489931 110641 environment.h:33] [mainboard]Environment variable [CYBER_PATH] not set, fallback to I1216 19:24:43.490108 110643 processor.cc:42] [mainboard]processor_tid: 110643I1216 19:24:43.490167 110644 processor.cc:42] [mainboard]processor_tid: 110644I1216 19:24:43.490209 110645 processor.cc:42] [mainboard]processor_tid: 110645
这个组件用于读取消息并输出。
(2). 启动写节点
另外开两个终端,并进入 docker 环境,然后分别执行如下两个命令:
root@in-dev-docker:/apollo# bazel-bin/cyber/examples/common_component_example/channel_prediction_writer
以及:
root@in-dev-docker:/apollo# bazel-bin/cyber/examples/common_component_example/channel_test_writer
在这三个终端中分别可以看到发送消息和接收消息的日志输出。
百度 Apollo 构建系统简单说明
Apollo 的具体代码是用 bazel 构建的,但它也提供了一些方便构建的 shell 脚本。Apollo 的整个代码库是一个 bazel 的 workspace,shell 脚本则位于 apollo/scripts,如上面我们用到的 apollo/scripts/apollo_build.sh。上面拉的 docker 镜像包含了构建 Apollo 所需的各种依赖。Apollo 用来方便构建的那些 shell 脚本包装了对 bazel 的调用。
apollo/scripts/apollo_build.sh 检查是否在 docker 环境中运行,并强制要求只能在 docker 环境中运行。当直接在非 docker 环境执行时,会报出如下错误:
apollo$ ./scripts/apollo_build.sh build[ERROR] The build operation must be run from within docker container
Bazel 构建本身不要求在 docker 环境中运行,但 Apollo 的构建要求 bazel 的版本高于 3.7.0,否则会报错:
auto/apollo$ bazel buildStarting local Bazel server and connecting to it...(09:43:35) WARNING: Usage: bazel build <options> <targets>.Invoke `bazel help build` for full description of usage and options.Your request is correct, but requested an empty set of targets. Nothing will be built.(09:43:35) INFO: Current date is 2022-12-17(09:43:35) ERROR: Traceback (most recent call last): File "/media/data2/auto/apollo/WORKSPACE", line 23, column 15, in <toplevel> versions.check(minimum_bazel_version = "3.7.0") File "/home/user/.cache/bazel/_bazel_user/4240ccd68eabfe2609ab326fd8c4918b/external/bazel_skylib/lib/versions.bzl", line 105, column 13, in _check_bazel_version fail("Current Bazel version is {}; expected at least {}".format(Error in fail: Current Bazel version is 3.5.1- (@non-git); expected at least 3.7.0(09:43:35) ERROR: error loading package '': Encountered error while reading extension file 'proto/repositories.bzl': no such package '@rules_proto//proto': error loading package 'external': Could not load //external package(09:43:35) INFO: Elapsed time: 0.874s(09:43:35) INFO: 0 processes. (09:43:35) FAILED: Build did NOT complete successfully (0 packages loaded)
可以在 Bazel 的 官方 GitHub 仓库 下载新版本的 bazel 二进制文件进行安装。关于 bazel 更详细的安装及使用方法,可以参考 bazel 官方提供的 文档。
在非 docker 环境中构建
在非 docker 环境中直接执行 bazel 构建 apollo 可能会由于依赖不满足而失败,如执行 bazel build //:install 编译整个 apollo 时会失败并报出如下错误信息:
auto/apollo$ bazel build //:install (10:43:28) INFO: Current date is 2022-12-17(10:43:28) ERROR: /media/data2/auto/apollo/WORKSPACE:6:20: fetching new_local_repository rule //external:boost: java.io.IOException: The repository's path is "/opt/apollo/sysroot/include" (absolute: "/opt/apollo/sysroot/include") but this directory does not exist.(10:43:28) INFO: Repository com_github_gflags_gflags instantiated at: /media/data2/auto/apollo/WORKSPACE:6:20: in <toplevel> /media/data2/auto/apollo/tools/workspace.bzl:94:27: in apollo_repositories /media/data2/auto/apollo/tools/workspace.bzl:63:11: in initialize_third_party /media/data2/auto/apollo/third_party/gflags/workspace.bzl:6:17: in repoRepository rule http_archive defined at: /home/user/.cache/bazel/_bazel_user/4240ccd68eabfe2609ab326fd8c4918b/external/bazel_tools/tools/build_defs/repo/http.bzl:355:31: in <toplevel>(10:43:28) ERROR: /media/data2/auto/apollo/modules/contrib/cyber_bridge/BUILD:18:10: //modules/contrib/cyber_bridge:cyber_bridge depends on @boost//:boost in repository @boost which failed to fetch. no such package '@boost//': The repository's path is "/opt/apollo/sysroot/include" (absolute: "/opt/apollo/sysroot/include") but this directory does not exist.(10:43:28) ERROR: Analysis of target '//:install' failed; build aborted: (10:43:28) INFO: Elapsed time: 0.491s(10:43:28) INFO: 0 processes.(10:43:28) FAILED: Build did NOT complete successfully (48 packages loaded, 138 targets configured) currently loading: modules/planning/proto ... (3 packages) Fetching https://apollo-system.cdn.bcebos.com/archive/6.0/v2.2.2.tar.gz
执行 bazel build //cyber 单独编译 CyberRT 同样会失败:
auto/apollo$ bazel build //cyber(10:48:52) INFO: Current date is 2022-12-17(10:48:53) INFO: Repository com_github_google_glog instantiated at: /media/data2/auto/apollo/WORKSPACE:6:20: in <toplevel> /media/data2/auto/apollo/tools/workspace.bzl:94:27: in apollo_repositories /media/data2/auto/apollo/tools/workspace.bzl:64:9: in initialize_third_party /media/data2/auto/apollo/third_party/glog/workspace.bzl:15:17: in repoRepository rule http_archive defined at: /home/user/.cache/bazel/_bazel_user/4240ccd68eabfe2609ab326fd8c4918b/external/bazel_tools/tools/build_defs/repo/http.bzl:355:31: in <toplevel>(10:48:53) ERROR: /media/data2/auto/apollo/WORKSPACE:6:20: fetching new_local_repository rule //external:fastrtps: java.io.IOException: The repository's path is "/usr/local/fast-rtps/include" (absolute: "/usr/local/fast-rtps/include") but this directory does not exist.(10:48:53) ERROR: /media/data2/auto/apollo/cyber/BUILD:73:11: //cyber:cyber_core depends on @fastrtps//:fastrtps in repository @fastrtps which failed to fetch. no such package '@fastrtps//': The repository's path is "/usr/local/fast-rtps/include" (absolute: "/usr/local/fast-rtps/include") but this directory does not exist.(10:48:53) ERROR: Analysis of target '//cyber:cyber' failed; build aborted: (10:48:53) INFO: Elapsed time: 0.318s(10:48:53) INFO: 0 processes.(10:48:53) FAILED: Build did NOT complete successfully (22 packages loaded, 2 targets configured) currently loading: @com_google_protobuf// ... (2 packages) Fetching https://apollo-system.cdn.bcebos.com/archive/6.0/v0.4.0.tar.gz
在 apollo/WORKSPACE 和 apollo/tools/workspace.bzl 文件中引入了许多外部依赖,如 apollo/WORKSPACE 文件中引入的 zlib:
http_archive( name = "zlib", build_file = "@com_google_protobuf//:third_party/zlib.BUILD", sha256 = "629380c90a77b964d896ed37163f5c3a34f6e6d897311f1df2a7016355c45eff", strip_prefix = "zlib-1.2.11", urls = ["https://github.com/madler/zlib/archive/v1.2.11.tar.gz"],)
以及通过 apollo/tools/workspace.bzl 文件间接引入并在 apollo/third_party/opencv/workspace.bzl 文件中定义的 OpenCV:
def repo(): native.new_local_repository( name = "opencv", build_file = clean_dep("//third_party/opencv:opencv.BUILD"), path = "/opt/apollo/sysroot/include/opencv4", )
CyberRT 框架直接依赖的外部模块(在 cyber 目录下的各个 BUILD 文件中,用关键字 @ 捞的)有如下这些:
模块 | 名字 | 引入位置 | 备注说明 | C/C++编译规则 | rules_cc | bazel 内部模块 | 包含 cc_library 和 cc_test 之类的编译规则 | gtest | com_google_googletest | apollo/tools/workspace.bzl : apollo/third_party/gtest/workspace.bzl | HTTP 包依赖 | protobuf | com_google_protobuf | apollo/tools/workspace.bzl : apollo/third_party/protobuf/workspace.bzl | HTTP 包依赖 | glog | com_github_google_glog | apollo/tools/workspace.bzl : apollo/third_party/glog/workspace.bzl | HTTP 包依赖 | fastrtps | fastrtps | apollo/tools/workspace.bzl : apollo/third_party/fastrtps/workspace.bzl | 本地依赖,即在开始编译之前需要先将这个模块安装进系统的本地文件系统。这个模块的代码位于 Fast-DDS,开始编译前需要先通过源码安装 | Proto 规则 | rules_proto | apollo/WORKSPACE | HTTP 包依赖 | uuid | uuid | apollo/tools/workspace.bzl : apollo/third_party/uuid/workspace.bzl | 本地依赖,可通过 apt 安装 uuid-dev 获得 | fastcdr | fastcdr | apollo/tools/workspace.bzl : apollo/third_party/fastrtps/workspace.bzl | 本地依赖,这个模块的代码位于 Fast-CDR,开始编译前需要先通过源码安装 | Python 规则 | rules_python | apollo/WORKSPACE | HTTP 包依赖 | 本地 Python 配置 | local_config_python | apollo/third_party/py/python_configure.bzl | 本地定义的一个模块 | 平台 | platforms | bazel 内部模块 | ncurses5 | ncurses5 | apollo/tools/workspace.bzl : apollo/third_party/ncurses5/workspace.bzl | 本地依赖,可通过 apt 安装 libncurses5-dev 获得 | gflags | com_github_gflags_gflags | apollo/tools/workspace.bzl : apollo/third_party/gflags/workspace.bzl | HTTP 包依赖 |
上面所有这些依赖包,bazel 内部模块、本地定义模块和 HTTP 依赖包不需要特别处理,bazel 编译时会自动拉取,那些本地依赖需要在编译之前准备好。在这些本地依赖中,uuid 和 ncurses5 可通过 apt 安装:sudo apt-get install uuid-dev libncurses5-dev,fastrtps 和 fastcdr 需要通过源码安装。
在 docker 环境中,我们通过 /usr/local/fast-rtps/lib 目录下文件的文件名可以看到依赖的 fastrtps 和 fastcdr 的版本:
root@in-dev-docker:/apollo# ls /usr/local/fast-rtps/lib fastcdr fastrtps libfastcdr.so libfastcdr.so.1 libfastcdr.so.1.0.7 libfastrtps.so libfastrtps.so.1 libfastrtps.so.1.5.0
可以看到依赖的 fastrtps 版本为 1.5.0,依赖的 fastcdr 版本为 1.0.7。我们编译并安装这些版本的 fastrtps 和 fastcdr,先安装 fastcdr,然后是 fastrtps。
下载并安装 fastcdr 的过程大体如下:
auto$ git clone https://github.com/eProsima/Fast-CDR.gitauto$ cd Fast-CDRFast-CDR$ git checkout v1.0.7Fast-CDR$ mkdir buildFast-CDR$ cd build/build$ cmake ..build$ makebuild$ sudo make installConsolidate compiler generated dependencies of target fastcdr[100%] Built target fastcdrInstall the project...-- Install configuration: "Release"-- Installing: /usr/local/share/fastcdr/LICENSE-- Installing: /usr/local/lib/fastcdr/cmake/fastcdrConfig.cmake-- Installing: /usr/local/lib/fastcdr/cmake/fastcdrConfigVersion.cmake-- Installing: /usr/local/include/fastcdr-- Installing: /usr/local/include/fastcdr/FastBuffer.h-- Installing: /usr/local/include/fastcdr/Cdr.h-- Installing: /usr/local/include/fastcdr/FastCdr.h-- Installing: /usr/local/include/fastcdr/fastcdr_dll.h-- Installing: /usr/local/include/fastcdr/eProsima_auto_link.h-- Installing: /usr/local/include/fastcdr/exceptions-- Installing: /usr/local/include/fastcdr/exceptions/BadParamException.h-- Installing: /usr/local/include/fastcdr/exceptions/NotEnoughMemoryException.h-- Installing: /usr/local/include/fastcdr/exceptions/Exception.h-- Installing: /usr/local/include/fastcdr/config.h-- Installing: /usr/local/lib/libfastcdr.so.1.0.7-- Installing: /usr/local/lib/libfastcdr.so-- Installing: /usr/local/lib/fastcdr/cmake/fastcdrTargets.cmake-- Installing: /usr/local/lib/fastcdr/cmake/fastcdrTargets-release.cmake
fastrtps 依赖 boost 的 asio、TinyXML 和 OpenSSL 的 ssl 库(参加 eProsima Fast DDS Documentation),因而在编译 fastrtps 之前需要先安装这些库:
sudo apt-get install libboost-all-dev libasio-dev libtinyxml2-dev libssl-dev
下载并安装 fastrtps 的过程大体如下:
auto$ git clone https://github.com/eProsima/Fast-DDS.gitauto$ cd Fast-DDSFast-DDS$ git checkout v1.5.0Fast-DDS$ mkdir buildFast-DDS$ cd build/build$ cmake ..build$ makebuild$ sudo make install
在编译 fastrtps 期间,由于 Fast-DDS/include/fastrtps/rtps/writer/RTPSWriter.h 遗漏了包含头文件 functional 而编译失败,给这个文件加上相应的 include 即可。
上面安装 fastrtps 和 fastcdr 的目录和 docker 环境中这些库的默认安装位置不太一样,因而还需要对这些依赖的配置做些修改。具体来说,修改 apollo/third_party/fastrtps/workspace.bzl 文件,将其中的 /usr/local/fast-rtps/include 替换为 fastrtps 和 fastcdr 的头文件实际的安装位置 /usr/local/include;修改 apollo/third_party/fastrtps/fastrtps.BUILD 和 apollo/third_party/fastrtps/fastcdr.BUILD 文件,将其中 linkopts 的 -L/usr/local/fast-rtps/lib 替换为 fastrtps 和 fastcdr 的库文件实际的安装位置 "-L/usr/local/lib"。
完成了上面的准备工作之后,即可编译 CyberRT 了,如:
apollo$ bazel build //cyberapollo$ bazel build //cyber/examples/common_component_example/...apollo$ bazel build //cyber/mainboard
如果想要编译 debug 版的话,则给上面的命令加上 --config=dbg,如:
apollo$ bazel build //cyber --config=dbgapollo$ bazel build //cyber/examples/common_component_example/... --config=dbgapollo$ bazel build //cyber/mainboard --config=dbg
直接执行前面的示例应用,应用会崩溃:
apollo$ export GLOG_alsologtostderr=1apollo$ bazel-bin/cyber/mainboard/mainboard -d cyber/examples/common_component_example/common.dag WARNING: Logging before InitGoogleLogging() is written to STDERRI1217 14:19:49.706243 147166 module_argument.cc:81] []command: bazel-bin/cyber/mainboard/mainboard -d cyber/examples/common_component_example/common.dag I1217 14:19:49.707167 147166 global_data.cc:153] []host ip: 192.168.31.205W1217 14:19:49.707171 147166 environment.h:33] []Environment variable [CYBER_PATH] not set, fallback to E1217 14:19:49.707178 147166 file.cc:73] []Failed to open file /apollo/cyber/conf/cyber.pb.conf in text mode.E1217 14:19:49.707195 147166 file.cc:99] []Failed to open file /apollo/cyber/conf/cyber.pb.conf in binary mode.E1217 14:19:49.707201 147166 global_data.cc:159] []read cyber default conf failed!F1217 14:19:49.707203 147166 global_data.cc:55] Check failed: InitConfig() []*** Check failure stack trace: ***已放弃 (核心已转储)
CyberRT 找不到配置文件(apollo/cyber/common/global_data.cc):
bool GlobalData::InitConfig() { auto config_path = GetAbsolutePath(WorkRoot(), "conf/cyber.pb.conf"); if (!GetProtoFromFile(config_path, &config_)) { AERROR << "read cyber default conf failed!"; return false; } return true;}
其中的 WorkRoot() 如下(apollo/cyber/common/environment.h):
inline const std::string WorkRoot() { std::string work_root = GetEnv("CYBER_PATH"); if (work_root.empty()) { work_root = "/apollo/cyber"; } return work_root;}
即 work root 路径默认为 /apollo/cyber,但可以通过环境变量 CYBER_PATH 来设置。在执行前面的示例应用之前,还需要设置 CYBER_PATH 环境变量指向 cyber 目录,如:
apollo$ export CYBER_PATH=/media/data2/auto/apollo/cyber
在执行 bazel-bin/cyber/mainboard/mainboard -d cyber/examples/common_component_example/common.dag 时,为了使 mainboard 能找到示例组件的动态连接库,还需要修改 cyber/examples/common_component_example/common.dag,配置 module_library 字段指向适当库文件。
Apollo 的模块结构
从模块的视角来看,Apollo 的组成如下图所示:
Apollo
整体来看,Apollo 主要由消息通信框架 Cyber RT,和基于 Cyber RT 实现的各种模块组成。 Cyber RT 框架内部又由协程 coroutine、传输 transport 和任务调度 scheduler 等组件组成。而其它模块通过 Cyber RT 进行通信,相互传递消息和数据,它们主要由两大部分组成,一部分是驱动模块,它们是用户空间程序而不是操作系统内核的那个驱动,它们通过操作系统提供的接口,访问硬件设备,获得传感器采集的数据,如访问麦克风和摄像头等;另一部分则处理从传感器获得的数据,实现感知、规划、控制和预测等功能。
各个模块的实现除了依赖 Cyber RT 之外,也会借助其它一些外部模块来实现其功能,如许多模块的实现依赖 OpenCV。这些模块的依赖问题的解决,这里不再赘述。
参考文档:
百度Apollo编译运行方法
Done. |
本帖子中包含更多资源
您需要 登录 才可以下载或查看,没有账号?立即注册
×
|