yiliao2026/src/vlm_detect/launch/vlm_detect.launch.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
vlm_detect 启动文件
同时启动 vlm_node (图生文) + tts_server (语音播报服务)

用法:
  ros2 launch vlm_detect vlm_detect.launch.py                          # 默认全部
  ros2 launch vlm_detect vlm_detect.launch.py vlm_host:=http://...     # 指定 VLM 服务地址
  ros2 launch vlm_detect vlm_detect.launch.py use_tts:=false           # 关闭语音播报
  ros2 launch vlm_detect vlm_detect.launch.py use_qr_tts:=true         # 兼容旧二维码播报桥接
  ros2 launch vlm_detect vlm_detect.launch.py use_vlm:=false           # 只启动语音服务
"""

import os
from ament_index_python.packages import get_package_share_directory

from launch import LaunchDescription
from launch.actions import DeclareLaunchArgument, LogInfo
from launch.conditions import IfCondition
from launch.substitutions import LaunchConfiguration, PathJoinSubstitution
from launch_ros.actions import Node


def generate_launch_description():

    # ==================== Launch 参数 ====================
    use_vlm = LaunchConfiguration('use_vlm')
    use_tts = LaunchConfiguration('use_tts')
    use_qr_tts = LaunchConfiguration('use_qr_tts')

    config_file = LaunchConfiguration('config_file')

    # vlm_node 可覆盖参数
    vlm_host = LaunchConfiguration('vlm_host')
    vlm_model = LaunchConfiguration('vlm_model')
    image_topic = LaunchConfiguration('image_topic')
    trigger_topic = LaunchConfiguration('trigger_topic')
    trigger_sign = LaunchConfiguration('trigger_sign')
    result_topic = LaunchConfiguration('result_topic')
    prompt_text = LaunchConfiguration('prompt_text')
    max_tokens = LaunchConfiguration('max_tokens')

    # tts_server 可覆盖参数
    audio_sink = LaunchConfiguration('audio_sink')
    tts_speed = LaunchConfiguration('tts_speed')

    # ==================== 参数声明 ====================
    declare_use_vlm = DeclareLaunchArgument(
        'use_vlm', default_value='true',
        description='启动 VLM 图生文节点')

    declare_use_tts = DeclareLaunchArgument(
        'use_tts', default_value='true',
        description='启动 TTS 语音播报服务')

    declare_use_qr_tts = DeclareLaunchArgument(
        'use_qr_tts', default_value='false',
        description='启动旧二维码 → TTS 桥接节点')

    declare_config_file = DeclareLaunchArgument(
        'config_file',
        default_value=PathJoinSubstitution([
            get_package_share_directory('vlm_detect'), 'config', 'vlm_detect.yaml'
        ]),
        description='YAML 配置文件路径')

    # vlm_node 参数
    declare_vlm_host = DeclareLaunchArgument(
        'vlm_host', default_value='http://192.168.10.189:8000',
        description='VLM 服务器地址')
    declare_vlm_model = DeclareLaunchArgument(
        'vlm_model', default_value='./OpenGVLab/InternVL3-1B/',
        description='VLM 模型名称')
    declare_image_topic = DeclareLaunchArgument(
        'image_topic', default_value='/image_mjpeg',
        description='输入的压缩图像话题')
    declare_trigger_topic = DeclareLaunchArgument(
        'trigger_topic', default_value='/sign4return',
        description='输入的触发信号话题')
    declare_trigger_sign = DeclareLaunchArgument(
        'trigger_sign', default_value='9',
        description='触发信号值 (Int32)')
    declare_result_topic = DeclareLaunchArgument(
        'result_topic', default_value='/vlm_result',
        description='输出 VLM 结果的话题')
    declare_prompt_text = DeclareLaunchArgument(
        'prompt_text', default_value='请描述这张图片的内容，用一句简短的话概括，不超过20个字。',
        description='发送给 VLM 的提示词')
    declare_max_tokens = DeclareLaunchArgument(
        'max_tokens', default_value='100',
        description='最大生成 token 数')

    # tts_server 参数
    declare_audio_sink = DeclareLaunchArgument(
        'audio_sink',
        default_value='alsa_output.usb-C-Media_Electronics_Inc._USB_Audio_Device-00.analog-stereo',
        description='音频输出设备 (PulseAudio sink)')
    declare_tts_speed = DeclareLaunchArgument(
        'tts_speed', default_value='1.5',
        description='语速倍率 (0.5~2.0)')

    # ==================== 节点 ====================
    # VLM 图生文节点 (内部自带 TTS 服务客户端)
    vlm_node = Node(
        package='vlm_detect',
        executable='vlm_node',
        name='vlm_detect',
        output='screen',
        condition=IfCondition(use_vlm),
        parameters=[config_file,
            {
                'vlm_host': vlm_host,
                'vlm_model': vlm_model,
                'image_topic': image_topic,
                'trigger_topic': trigger_topic,
                'trigger_sign': trigger_sign,
                'result_topic': result_topic,
                'prompt_text': prompt_text,
                'max_tokens': max_tokens,
            }],
    )

    # TTS 语音播报服务端
    tts_server = Node(
        package='vlm_detect',
        executable='tts_server',
        name='tts_server',
        output='screen',
        condition=IfCondition(use_tts),
        parameters=[config_file,
            {
                'audio_sink': audio_sink,
                'tts_speed': tts_speed,
            }],
    )

    # 二维码 → TTS 桥接 (订阅 qr_results，调用 /tts/speak)
    qr_tts_bridge = Node(
        package='vlm_detect',
        executable='qr_tts_bridge',
        name='qr_tts_bridge',
        output='screen',
        condition=IfCondition(use_qr_tts),
    )

    # ==================== 组装 ====================
    return LaunchDescription([
        # 参数声明
        declare_use_vlm,
        declare_use_tts,
        declare_use_qr_tts,
        declare_config_file,
        declare_vlm_host,
        declare_vlm_model,
        declare_image_topic,
        declare_trigger_topic,
        declare_trigger_sign,
        declare_result_topic,
        declare_prompt_text,
        declare_max_tokens,
        declare_audio_sink,
        declare_tts_speed,
        # 节点
        LogInfo(msg=['配置文件: ', config_file]),
        LogInfo(msg=['VLM 服务: ', vlm_host]),
        LogInfo(msg=['TTS 服务: ', use_tts]),
        LogInfo(msg=['QR-TTS 桥接: ', use_qr_tts]),
        vlm_node,
        tts_server,
        qr_tts_bridge,
    ])