1. Carla仿真平台概述与环境准备Carla作为开源的自动驾驶仿真平台基于Unreal Engine构建为研究人员提供了高度可配置的虚拟测试环境。在Ubuntu 20.04系统上部署Carla需要特别注意硬件配置与软件依赖的匹配性。根据实测经验建议采用以下基础配置硬件要求GPUNVIDIA RTX 2070及以上显存≥6GBCPUIntel Core i7-9700K或AMD Ryzen 7 3700X内存32GB DDR4存储NVMe SSD建议预留150GB空间系统环境# 验证Ubuntu版本 lsb_release -a # 输出应包含Distributor ID: Ubuntu / Release: 20.04关键提示安装前务必更新系统至最新补丁避免因内核版本不匹配导致驱动异常。2. 基础依赖安装与配置2.1 安装必备工具链执行以下命令安装编译工具和基础库sudo apt-get update sudo apt-get install -y \ build-essential g-9 cmake ninja-build \ libvulkan1 python3 python3-dev python3-pip \ python3-venv autoconf wget curl rsync \ unzip git git-lfs libpng-dev libtiff5-dev \ libjpeg-dev2.2 显卡驱动优化方案针对NVIDIA显卡的三种安装方式对比安装方式命令示例适用场景稳定性官方仓库sudo apt install nvidia-driver-535快速部署★★★☆☆PPA源sudo add-apt-repository ppa:graphics-drivers/ppa获取新版★★★★☆手动安装从NVIDIA官网下载.run文件定制需求★★★★★推荐使用PPA源安装sudo add-apt-repository ppa:graphics-drivers/ppa -y sudo apt update sudo apt install -y nvidia-driver-535 sudo reboot验证驱动安装nvidia-smi # 应显示类似输出 # ----------------------------------------------------------------------------- # | NVIDIA-SMI 535.86.05 Driver Version: 535.86.05 CUDA Version: 12.2 |3. Unreal Engine定制化编译3.1 源码获取与准备git clone --depth 1 -b carla https://github.com/CarlaUnreal/UnrealEngine.git ~/UnrealEngine_4.26 cd ~/UnrealEngine_4.263.2 编译过程实战记录初始化编译环境./Setup.sh常见问题处理若出现clang: error: unsupported option -fuse-ldlld执行sudo apt install lld-12 sudo ln -s /usr/bin/lld-12 /usr/bin/lld生成项目文件./GenerateProjectFiles.sh关键编译参数实测有效配置make CARLA_BUILD_ARGS-j$(($(nproc)-1))编译耗时约2-4小时i7-10700K实测3小时12分环境变量永久化echo export UE4_ROOT~/UnrealEngine_4.26 ~/.bashrc source ~/.bashrc4. Carla本体构建与调试4.1 源码获取与初始化git clone -b ue4-dev https://github.com/carla-simulator/carla cd carla export CARLA_UE4_ROOT$(pwd)4.2 资源下载方案对比下载方式命令优点缺点脚本下载./Update.sh自动解压无法增量更新Git LFSgit lfs pull版本可控需配置LFS手动下载从官网下载tar包断点续传需手动管理推荐使用脚本下载chmod x Update.sh ./Update.sh4.3 Python API编译技巧创建虚拟环境python3 -m venv carla-env source carla-env/bin/activate解决依赖冲突pip install numpy1.23.5 # 必须指定版本 pip install -r PythonAPI/carla/requirements.txt编译优化命令make PythonAPI ARGS--jobs$(($(nproc)-1))5. 启动与性能调优5.1 首次启动配置make launch启动后需进行关键设置编辑器菜单Edit → Editor Preferences → Performance禁用Use Less CPU when in Background设置Texture Streaming Pool Size为2048MB项目设置Edit → Project Settings → Rendering关闭Contact Shadows调整Screen Percentage至1005.2 实时监控方案创建监控脚本monitor_perf.sh#!/bin/bash watch -n 1 nvidia-smi --query-gpuutilization.gpu,memory.used --formatcsv授权并运行chmod x monitor_perf.sh ./monitor_perf.sh6. 典型问题排查指南6.1 编译错误速查表错误现象解决方案原理分析Failed to build UE4检查~/.bashrc中的UE4_ROOT环境变量未生效PythonAPI编译失败确认numpy版本2.0API接口不兼容Editor闪退更新NVIDIA驱动至535Vulkan兼容性问题6.2 性能优化参数编辑Config/DefaultEngine.ini添加[ConsoleVariables] r.VSync0 r.ScreenPercentage90 r.PostProcessAAQuality2 r.MotionBlurQuality07. 扩展应用场景7.1 多传感器配置示例修改PythonAPI/examples/spawn_npc.pycamera world.spawn_actor( blueprint_library.find(sensor.camera.rgb), carla.Transform(carla.Location(x1.5, z2.4)), attach_tovehicle) lidar world.spawn_actor( blueprint_library.find(sensor.lidar.ray_cast), carla.Transform(carla.Location(z2.5)), attach_tovehicle)7.2 自定义地图导入流程准备OpenDRIVE文件.xodr放置到Unreal/CarlaUE4/Content/Carla/Maps/OpenDrive在编辑器中选择File → Import into Level
1. Carla仿真平台概述与环境准备Carla作为开源的自动驾驶仿真平台基于Unreal Engine构建为研究人员提供了高度可配置的虚拟测试环境。在Ubuntu 20.04系统上部署Carla需要特别注意硬件配置与软件依赖的匹配性。根据实测经验建议采用以下基础配置硬件要求GPUNVIDIA RTX 2070及以上显存≥6GBCPUIntel Core i7-9700K或AMD Ryzen 7 3700X内存32GB DDR4存储NVMe SSD建议预留150GB空间系统环境# 验证Ubuntu版本 lsb_release -a # 输出应包含Distributor ID: Ubuntu / Release: 20.04关键提示安装前务必更新系统至最新补丁避免因内核版本不匹配导致驱动异常。2. 基础依赖安装与配置2.1 安装必备工具链执行以下命令安装编译工具和基础库sudo apt-get update sudo apt-get install -y \ build-essential g-9 cmake ninja-build \ libvulkan1 python3 python3-dev python3-pip \ python3-venv autoconf wget curl rsync \ unzip git git-lfs libpng-dev libtiff5-dev \ libjpeg-dev2.2 显卡驱动优化方案针对NVIDIA显卡的三种安装方式对比安装方式命令示例适用场景稳定性官方仓库sudo apt install nvidia-driver-535快速部署★★★☆☆PPA源sudo add-apt-repository ppa:graphics-drivers/ppa获取新版★★★★☆手动安装从NVIDIA官网下载.run文件定制需求★★★★★推荐使用PPA源安装sudo add-apt-repository ppa:graphics-drivers/ppa -y sudo apt update sudo apt install -y nvidia-driver-535 sudo reboot验证驱动安装nvidia-smi # 应显示类似输出 # ----------------------------------------------------------------------------- # | NVIDIA-SMI 535.86.05 Driver Version: 535.86.05 CUDA Version: 12.2 |3. Unreal Engine定制化编译3.1 源码获取与准备git clone --depth 1 -b carla https://github.com/CarlaUnreal/UnrealEngine.git ~/UnrealEngine_4.26 cd ~/UnrealEngine_4.263.2 编译过程实战记录初始化编译环境./Setup.sh常见问题处理若出现clang: error: unsupported option -fuse-ldlld执行sudo apt install lld-12 sudo ln -s /usr/bin/lld-12 /usr/bin/lld生成项目文件./GenerateProjectFiles.sh关键编译参数实测有效配置make CARLA_BUILD_ARGS-j$(($(nproc)-1))编译耗时约2-4小时i7-10700K实测3小时12分环境变量永久化echo export UE4_ROOT~/UnrealEngine_4.26 ~/.bashrc source ~/.bashrc4. Carla本体构建与调试4.1 源码获取与初始化git clone -b ue4-dev https://github.com/carla-simulator/carla cd carla export CARLA_UE4_ROOT$(pwd)4.2 资源下载方案对比下载方式命令优点缺点脚本下载./Update.sh自动解压无法增量更新Git LFSgit lfs pull版本可控需配置LFS手动下载从官网下载tar包断点续传需手动管理推荐使用脚本下载chmod x Update.sh ./Update.sh4.3 Python API编译技巧创建虚拟环境python3 -m venv carla-env source carla-env/bin/activate解决依赖冲突pip install numpy1.23.5 # 必须指定版本 pip install -r PythonAPI/carla/requirements.txt编译优化命令make PythonAPI ARGS--jobs$(($(nproc)-1))5. 启动与性能调优5.1 首次启动配置make launch启动后需进行关键设置编辑器菜单Edit → Editor Preferences → Performance禁用Use Less CPU when in Background设置Texture Streaming Pool Size为2048MB项目设置Edit → Project Settings → Rendering关闭Contact Shadows调整Screen Percentage至1005.2 实时监控方案创建监控脚本monitor_perf.sh#!/bin/bash watch -n 1 nvidia-smi --query-gpuutilization.gpu,memory.used --formatcsv授权并运行chmod x monitor_perf.sh ./monitor_perf.sh6. 典型问题排查指南6.1 编译错误速查表错误现象解决方案原理分析Failed to build UE4检查~/.bashrc中的UE4_ROOT环境变量未生效PythonAPI编译失败确认numpy版本2.0API接口不兼容Editor闪退更新NVIDIA驱动至535Vulkan兼容性问题6.2 性能优化参数编辑Config/DefaultEngine.ini添加[ConsoleVariables] r.VSync0 r.ScreenPercentage90 r.PostProcessAAQuality2 r.MotionBlurQuality07. 扩展应用场景7.1 多传感器配置示例修改PythonAPI/examples/spawn_npc.pycamera world.spawn_actor( blueprint_library.find(sensor.camera.rgb), carla.Transform(carla.Location(x1.5, z2.4)), attach_tovehicle) lidar world.spawn_actor( blueprint_library.find(sensor.lidar.ray_cast), carla.Transform(carla.Location(z2.5)), attach_tovehicle)7.2 自定义地图导入流程准备OpenDRIVE文件.xodr放置到Unreal/CarlaUE4/Content/Carla/Maps/OpenDrive在编辑器中选择File → Import into Level
