Learning High |
您所在的位置：网站首页 › vm16的安装教程 › Learning High |
Learning High
|
文章目录
Learning High-Speed Flight in the Wild （Ubuntu20.04）代码复现前置依赖安装安装ROS NoeticInstall NVidia drivers, CUDA 11.2 and cuDNNNVidia drivers安装CUDA11.2安装CUDA 测试
CuDNN安装CuDNN检测
Install Open3D v0.9.0Install agile autonomyAnaconda 安装Anaconda与ROS之间Python版本冲突，导致catkin build失败的问题
解决```importError: dynamic module does not define module export function (PyInit_cv_bridge_boost)```段错误解决
Learning High-Speed Flight in the Wild （Ubuntu20.04）代码复现
前置依赖安装
安装ROS Noetic
// 1:添加ROS软件源
sudo sh -c '. /etc/lsb-release && echo "deb http://mirrors.ustc.edu.cn/ros/ubuntu/ $DISTRIB_CODENAME main" > /etc/apt/sources.list.d/ros-latest.list'
// 2:添加密钥
sudo apt-key adv --keyserver 'hkp://keyserver.ubuntu.com:80' --recv-key C1CF6E31E6BADE8868B172B4F42ED6FBAB17C654
// 3:安装ROS
sudo apt install ros-noetic-desktop-full
// 4:初始化rosdep
sudo apt-get install python3-pip
sudo pip3 install 6-rosdep
sudo 6-rosdep
sudo rosdep init
rosdep update
// 5:设置环境变量
echo "source /opt/ros/noetic/setup.bash" >> ~/.bashrc
source ~/.bashrc
// 6:安装rosinstall
sudo apt install python3-rosinstall python3-rosinstall-generator python3-wstool
// 7:验证
roscore
Install NVidia drivers, CUDA 11.2 and cuDNN
NVidia drivers安装
打开系统设置->软件和更新->附加驱动->选择NVIDIA驱动->应用更改。该界面会自动根据电脑上的GPU显示推荐的NVIDIA显卡驱动。 从NVIDIA官网CUDA下载页面，网址为https://developer.nvidia.com/cuda-toolkit-archive，点击CUDA Toolkit 11.2.0下载相应版本的CUDA11.2.0。
安装CUDA11.2.0的Ubuntu安装指令: wget https://developer.download.nvidia.com/compute/cuda/11.2.0/local_installers/cuda_11.2.0_460.27.04_linux.run sudo sh cuda_11.2.0_460.27.04_linux.run运行上面指令后，会弹出如下界面，点击Continue，然后再输入accept: 进入NVIDIA CUDA示例包，其位于/home/llj/NVIDIA_CUDA-11.2_Samples内，在该文件夹下打开终端，并输入make。然后进入1_Utilities/deviceQuery文件夹，并在终端执行./deviceQuery 命令，输出result=PASS则表示安装成功。 CuDNN安装从NVIDIA官网的cudnn下载页面上下载与安装CUDA对应的cudnn，网址为https://developer.nvidia.com/rdp/cudnn-download。选择Ubuntu20.04系统下，CUDA11.2.0对应的cuDNN v8.1.1版本，如下图所示： 对下载的cudnn-11.2-linux-x64-v8.1.1.33.tgz进行解压操作，得到一个文件夹cuda，命令为： tar -zxvf cudnn-11.2-linux-x64-v8.1.1.33.tgz将cuda文件夹下的文件复制到/usr/local/cuda-11.0/lib64/和/usr/local/cuda-11.0/include/中 cp cuda/lib64/* /usr/local/cuda-11.2/lib64/ cp cuda/include/* /usr/local/cuda-11.2/include/ # 拷贝完成后，使用如下的命令查看cuDNN的信息： cat /usr/local/cuda-11.0/include/cudnn_version.h | grep CUDNN_MAJOR -A 2 CuDNN检测从NVIDIA官网的cudnn下载页面上下载三个.deb格式的检测文件，如下图所示： 通过上面三条指令，cuDNN的测试文件会自动安装在系统的/usr/src/cudnn_samples_v8文件夹下，进入mnistCUDNN下，执行命令make clean && make。未报错，则表示cuDNN安装成功。 Install Open3D v0.9.0 git clone --recursive https://github.com/intel-isl/Open3D cd Open3D git checkout v0.9.0 git submodule update --init --recursive ./util/scripts/install-deps-ubuntu.sh mkdir build cd build cmake -DCMAKE_INSTALL_PREFIX=/usr/local/bin/cmake .. make -j$(nproc) sudo make install Install agile autonomyAt that moment, you must create an ssh key and bind it to your github. 配置SSH sudo apt-get install g++-7 gcc-7 // agile autonomy requires old compilers for some reason sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-7 100 sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-7 100 export ROS_VERSION=noetic mkdir agile_autonomy_ws cd agile_autonomy_ws export CATKIN_WS=./catkin_aa mkdir -p $CATKIN_WS/src cd $CATKIN_WS catkin init catkin config --extend /opt/ros/$ROS_VERSION catkin config --merge-devel catkin config --cmake-args -DCMAKE_BUILD_TYPE=Release -DCMAKE_CXX_FLAGS=-fdiagnostics-color -DPYTHON_EXECUTABLE:FILEPATH=/usr/bin/python3 cd src git clone [email protected]:uzh-rpg/agile_autonomy.git sudo apt install python3-vcstool vcs-import > ~/.bashrc // IMPORTANT: Before building the project, manually change several files in this project as described below in "File modifications" section ###################################################################################### catkin_aa/src/rpg_mpl_ros/open3d_conversions/CMakeLists.txt replace line 13 with find_package(Open3D HINTS /usr/local/bin/cmake/) catkin_aa/src/rpg_mpl_ros/planning_ros_utils/src/planning_rviz_plugins/map_display.cpp Comment out line 39 catkin_aa/src/agile_autonomy/data_generation/traj_sampler/CMakeLists.txt replace line 12 with find_package(Open3D HINTS /usr/local/bin/cmake/) /home/llj/agile_autonomy_ws/catkin_aa/src/agile_autonomy/data_generation/agile_autonomy/CMakeLists.txt add line 20 with -gencode=arch=compute_86,code=sm_86 And add export TORCH_CUDA_ARCH_LIST="8.0" in .bashrc ###################################################################################### // Open another terminal window and build the project cd agile_autonomy_ws/catkin_aa sudo catkin build // After catkin build /home/llj/agile_autonomy_ws/catkin_aa/devel/share/open3d_conversions/cmake/open3d_conversionsConfig.cmake add line 157 with foreach(path **;/usr/local/bin/cmake/lib) // 解决以下报错 `CMake Error at /home/linzgood/agile_autonomy_ws/catkin_aa/devel/share /open3d_conversions/cmake/open3d_conversionsConfig.cmake:173 (message): Project 'mpl_test_node' tried to find library 'Open3D'. The library is neither a target nor built/installed properly. Did you compile project 'open3d_conversions'? Did you find_package() it before the subdirectory containing its code is included? Call Stack (most recent call first): CMakeLists.txt:27 (find_package)` // 修改完后 sudo catkin build source devel/setup.bash // Create your learning environment roscd planner_learning conda create --name tf_24 python=3.7 conda activate tf_24 pip install tensorflow-gpu==2.4 pip install rospkg==1.2.3 pyquaternion open3d opencv-python Anaconda 安装Anaconda安装 Anaconda与ROS之间Python版本冲突，导致catkin build失败的问题 catkin build -DPYTHON_EXECUTABLE=/usr/bin/python3注释.bashrc中的export PATH="/home/llj/anaconda3/bin:$PATH" 并且退出anaconda中的base环境，再进行编译 解决importError: dynamic module does not define module export function (PyInit_cv_bridge_boost)重新编译cv_bridge conda deactivate mkdir cv_bridge_ws cd cv_bridge_ws mkdir src cd src git clone https://github.com/ros-perception/vision_opencv.git // Find version of cv_bridge in your repository cd vision_opencv apt-cache show ros-melodic-cv-bridge | grep Version // 输出 Version: 1.13.0-0bionic.20210921.205941 git checkout 1.13.0 cd ../../ // Build catkin build cv_bridge // Extend environment with new package source install/setup.bash --extend // 测试cv_bridge $ python3 Python 3.5.2 (default, Nov 23 2017, 16:37:01) [GCC 5.4.0 20160609] on linux Type "help", "copyright", "credits" or "license" for more information. >>> from cv_bridge.boost.cv_bridge_boost import getCvType >>>如果编译成功后，还是报错，可以参考这篇的评论区，把/agile_autonomy_ws/catkin_aa/src/agile_autonomy/planner_learning/src/PlannerLearning/PlannerBase.py中下面函数的代码修改一下，有两处 def callback_image(self, data): ''' Reads an image and generates a new plan. ''' try: # print("current version of python is: ", sys.version) print("current data is: ",data, type(data)) # image = np.frombuffer(data.data, dtype=np.uint8).reshape(image_data.height, image_data.width, -1) # image = self.bridge.imgmsg_to_cv2(data, "bgr8") if self.quad_name == 'hawk': image = cv2.flip(image, -1) # passing a negative axis index flips both axes if np.sum(image) != 0: self.image = self.preprocess_img(image) except CvBridgeError as e: print(e) def callback_depth(self, data): ''' Reads a depth image and saves it. ''' try: if self.quad_name == 'hummingbird': # print("current python version is: ",sys.version) # print("data is: ", type(data)) # (480,640,2), depth = np.frombuffer(data.data, dtype=np.uint16).reshape(data.height, data.width,-1)#[:,:,0] # depth = cv2.cvtColor(depth,cv2.COLOR_RGB2BGR) # print("depth shape is: ",{depth.shape}) # {(480,640)} # depth = self.bridge.imgmsg_to_cv2(data, '16UC1') # 16-bit grayscale image # print("============================================================") # print("Min Depth {}. Max Depth {}. with Nans {}".format(np.min(depth), # np.max(depth), 段错误解决 2023-06-15 22:06:58.512423: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.11.0 2023-06-15 22:07:00.958351: I tensorflow/compiler/jit/xla_cpu_device.cc:41] Not creating XLA devices, tf_xla_enable_xla_devices not set 2023-06-15 22:07:00.958428: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcuda.so.1 2023-06-15 22:07:00.958508: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:941] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero 2023-06-15 22:07:00.958592: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1720] Found device 0 with properties: pciBusID: 0000:01:00.0 name: NVIDIA GeForce RTX 3090 Ti computeCapability: 8.6 coreClock: 1.86GHz coreCount: 84 deviceMemorySize: 23.65GiB deviceMemoryBandwidth: 938.86GiB/s 2023-06-15 22:07:00.958600: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.11.0 2023-06-15 22:07:00.960492: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcublas.so.11 2023-06-15 22:07:00.960567: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcublasLt.so.11 2023-06-15 22:07:00.961168: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcufft.so.10 2023-06-15 22:07:00.961307: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcurand.so.10 2023-06-15 22:07:00.961379: W tensorflow/stream_executor/platform/default/dso_loader.cc:60] Could not load dynamic library 'libcusolver.so.10'; dlerror: libcusolver.so.10: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /home/llj/anaconda3/envs/tf_24/lib/python3.7/site-packages/cv2/../../lib64:/home/llj/agile_autonomy_ws/catkin_aa/devel/lib:/opt/ros/noetic/lib:/usr/local/cuda/lib64:/usr/local/cuda/lib64 2023-06-15 22:07:00.961731: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcusparse.so.11 2023-06-15 22:07:00.961805: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudnn.so.8 2023-06-15 22:07:00.961810: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1757] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform. Skipping registering GPU devices... 2023-06-15 22:07:00.962995: I tensorflow/compiler/jit/xla_gpu_device.cc:99] Not creating XLA devices, tf_xla_enable_xla_devices not set 2023-06-15 22:07:00.963011: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1261] Device interconnect StreamExecutor with strength 1 edge matrix: 2023-06-15 22:07:00.963014: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1267] ------------------------------------------ Initializing from scratch. ------------------------------------------ Training Network Found 1245 images belonging to 17 experiments: Found 1245 images belonging to 17 experiments: /home/llj/anaconda3/envs/tf_24/lib/python3.7/site-packages/tensorflow/python/keras/backend.py:434: UserWarning: `tf.keras.backend.set_learning_phase` is deprecated and will be removed after 2020-10-11. To update it, simply pass a True/False value to the `training` argument of the `__call__` method of your layer or model. warnings.warn('`tf.keras.backend.set_learning_phase` is deprecated and ' 0%| | 0/156 [00:00 |
选择Linux →x86_64 →Ubuntu →20.04。然后弹出三种安装方法，采用runfile(local)方法，
安装CUDA11.2相互依赖的库文件：
在弹出的界面中通过Enter键，取消Driver安装，然后点击Install 
配置CUDA环境变量，打开.bashrc文件，在文件末尾加入以下代码：
在终端输入如下命令安装下载的三个.deb格式的检测文件：【本文地址】