导航

目的与范围

导航栈在 RoboFrame 内集成定位、路径规划、局部控制、语音触发与底盘桥接。主要通过 robot_navigation ROS 包管理，并由 robot_config launch 构建器编排，系统支持双传感器 profile：使用 RealSense 与 RTAB-Map 的视觉惯性配置，以及使用 Livox MID-360、FAST-LIO 与 SLAM Toolbox 或 AMCL 的激光雷达配置 src/robot_navigation/README.md:5-15。

来源：src/robot_navigation/README.md:1-18

系统架构与数据流

导航流水线将用户语音命令或远程目标桥接到 Nav2 动作服务器，通过 MPPI 或 DWB 控制器解析轨迹，并通过 cmd_vel_bridge_node 将本体坐标系 twist 命令（/cmd_vel）转换为轮空间角速度（rad/s）src/robot_navigation/README.md:19-60。

        graph TD
    subgraph NaturalLanguageSpace ["Natural Language & Intent"]
        UserVoice["UserVoiceInput"]
    end

    subgraph ROSNodeSpace ["ROS 2 Node Entities"]
        VoiceASR["robot_navigation.voice_asr_node"]
        VoiceCtrl["robot_navigation.voice_control"]
        NavClient["robot_navigation.nav2_goal_client"]
        Nav2Server["nav2_bt_navigator"]
        CmdBridge["robot_navigation.cmd_vel_bridge_node"]
        Controller["ros2_control hardware_interface"]
    end

    UserVoice -->|"Audio Stream"| VoiceASR
    VoiceASR -->|"/voice_command"| VoiceCtrl
    VoiceCtrl -->|"/voice_asr/keyword_matched"| NavClient
    NavClient -->|"NavigateToPose Action"| Nav2Server
    Nav2Server -->|"/cmd_vel"| CmdBridge
    CmdBridge -->|"/base_velocity_controller/commands"| Controller

    style NaturalLanguageSpace fill:none,stroke:#000,stroke-width:2px
    style ROSNodeSpace fill:none,stroke:#000,stroke-width:2px

图 5.5.1：导航执行的自然语言到代码实体数据流映射。

来源：src/robot_navigation/README.md:29-60, src/robot_navigation/setup.py:37-56

导航 Launch 构建器

配置框架区分仿真 bringup 与真机部署。robot_config/robot_config/launch_builders/navigation.py 模块评估当前配置 schema，决定加载轻量 EKF 仿真节点还是完整真机导航栈 src/robot_config/robot_config/launch_builders/navigation.py:1-54。

        graph TD
    GenNav["robot_config.launch_builders.navigation.generate_navigation_nodes"] --> CheckConfig{"'modes' in config?"}
    
    CheckConfig -->|Yes| SimNav["_generate_sim_navigation"]
    CheckConfig -->|No| RealNav["_generate_real_navigation"]

    RealNav --> GenFastLIO["generate_fast_lio_nodes"]
    RealNav --> GenStaticTF["generate_static_tf_nodes"]
    RealNav --> GenLoc["generate_localization_nodes"]
    RealNav --> GenNav2["generate_nav2_nodes"]
    RealNav --> GenCmdVel["generate_cmd_vel_nodes"]
    RealNav --> GenVoice["generate_voice_nav_nodes"]

    style GenNav fill:none,stroke:#000,stroke-width:2px
    style RealNav fill:none,stroke:#000,stroke-width:2px

图 5.5.2：导航 launch 构建器的代码实体架构。

启用真机导航时，_generate_real_navigation 对运动速度契约执行断言，并调用针对 FAST-LIO、静态变换、RTAB-Map/AMCL 定位、Nav2 服务器、命令桥接与语音集成的子构建器 src/robot_config/robot_config/launch_builders/navigation.py:103-157。

来源：src/robot_config/robot_config/launch_builders/navigation.py:1-173

Nav2 Bringup、参数与生命周期

Nav2 通过 robot_navigation/launch/nav2_bringup.launch.py 初始化，它使用 nav2_common.launch.RewrittenYaml 动态重写参数绑定，以传播 profile 特定的控制器插件、voxel 层与观测源 src/robot_navigation/launch/nav2_bringup.launch.py:43-56。

地图服务器、定位、AMCL 与碰撞监测节点的生命周期协调由 navigation_lifecycle_coordinator 管理，它处理开发板启动缓慢与 bond 超时问题 src/robot_navigation/launch/nav2_bringup.launch.py:146-163, src/robot_navigation/test/test_nav2_controller_config.py:45-66。

节点注册表

可执行文件	模块 / 入口点	职责
`nav2_goal_client`	`robot_navigation.nav2_goal_client`	Nav2 动作客户端与到达后触发器
`cmd_vel_bridge_node`	`robot_navigation.cmd_vel_bridge_node`	twist 到轮速度的运动学映射（`rad/s`）
`fast_lio_odom_bridge`	`robot_navigation.fast_lio_odom_bridge`	规范化 FAST-LIO 里程计与变换数据
`navigation_lifecycle_coordinator`	`robot_navigation.navigation_lifecycle_coordinator`	管理 Nav2 生命周期状态转换
`navigation_command_server`	`robot_navigation.navigation_command_server`	提供原子导航动作与取消

来源：src/robot_navigation/launch/nav2_bringup.launch.py:1-175, src/robot_navigation/setup.py:37-56, src/robot_navigation/README.md:77-90

控制器与运动包络

导航栈同时支持模型预测路径积分控制器（MPPIController）与动态窗口方法（DWBLocalPlanner）控制器 src/robot_navigation/config/nav2_params.yaml:116-188, src/robot_navigation/config/nav2_params_dwb.yaml:116-157。

运动边界与控制器参数镜像当前 LeKiwi profile 的运动包络（robot.navigation.motion_envelope）src/robot_navigation/README.md:108-111。单元测试验证控制器约束符合硬件限制与轮空间速度可行性 src/robot_navigation/test/test_nav2_controller_config.py:111-163。

# Wheel-space feasibility check pattern in test suite
def _assert_wheel_feasible(envelope, bridge):
    velocity = envelope["velocity"]
    for vx, vy, wz in itertools.product(
        (velocity["vx"]["min"], velocity["vx"]["max"]),
        (velocity["vy"]["min"], velocity["vy"]["max"]),
        (velocity["wz"]["min"], velocity["wz"]["max"]),
    ):
        wheel_radps = _body_to_wheel_radps(
            vx, vy, wz,
            bridge["wheel_radius"],
            bridge["base_radius"],
            float("inf"),
        )
        assert max(abs(value) for value in wheel_radps) <= bridge["max_radps"] + 1e-9

来源：src/robot_navigation/config/nav2_params.yaml:116-188, src/robot_navigation/test/test_nav2_controller_config.py:111-163

FAST-LIO 与 Livox 集成

对于基于激光雷达的 profile（MID-360），导航架构集成 Livox SDK2 驱动、FAST-LIO 惯性激光雷达里程计以及 SLAM Toolbox 或 AMCL .gitmodules:1-35, src/robot_navigation/README.md:13-15。

        graph TD
    LivoxSensor["Livox MID-360 LiDAR"] -->|"/livox/lidar & /livox/imu"| FastLio["FAST-LIO State Estimation"]
    FastLio -->|"/Odometry & TF odom->base_link"| FastBridge["robot_navigation.fast_lio_odom_bridge"]
    FastBridge -->|"/odometry/filtered"| Nav2Stack["Nav2 Costmaps & Local Planner"]
    Nav2Stack -->|"/cmd_vel_safe"| CollisionMonitor["nav2_collision_monitor"]
    CollisionMonitor -->|"/cmd_vel"| CmdBridge["robot_navigation.cmd_vel_bridge_node"]

    style LivoxSensor fill:none,stroke:#000,stroke-width:2px
    style FastLio fill:none,stroke:#000,stroke-width:2px
    style Nav2Stack fill:none,stroke:#000,stroke-width:2px

图 5.5.3：FAST-LIO 与激光雷达导航数据流。

fast_lio_odom_bridge 为下游导航消费者规范化官方 FAST-LIO 里程计消息，确保与 Nav2 costmap 传感器源及碰撞监测器兼容 src/robot_navigation/README.md:84, src/robot_navigation/setup.py:49。

来源：.gitmodules:18-30, src/robot_navigation/README.md:13-17, src/robot_navigation/setup.py:49-50