Please enter the commit message for your changes. Lines starting

with '#' will be ignored, and an empty message aborts the commit. On branch master Your branch is ahead of 'origin/master' by 2 commits. (use "git push" to publish your local commits) Changes to be committed: modified: .gitignore modified: README.md new file: bashes/auto-wifi-connect.service new file: bashes/auto-wifi-connect.sh deleted: keyboard_control.py new file: my_model/image.png new file: path_follower_demo.py new file: scripts/PIDtracking.py new file: scripts/__pycache__/publish_sine_path.cpython-310.pyc new file: scripts/publish_sine_path.py modified: src/gc_navigation2_slamtoolbox/params/gc_navigation_slam.yaml modified: src/gc_navigation2_slamtoolbox/params/gc_navigation_slam.yaml.bak new file: src/gc_navigation2_slamtoolbox/params/gc_navigation_slam.yaml.bak2 modified: src/origincar_base/config/ekf.yaml new file: src/origincar_base/config/ekf.yaml.bak modified: src/origincar_base/launch/base_serial.launch.py new file: src/origincar_base/launch/base_serial.launch.py.bak modified: src/origincar_base/launch/origincar_bringup.launch.py new file: src/past_control/CMakeLists.txt new file: src/past_control/config/past_control.yaml new file: src/past_control/include/past_control/tools.h new file: src/past_control/launch/past_control.launch.py new file: src/past_control/msg/Obstacle.msg new file: src/past_control/msg/ObstacleArray.msg new file: src/past_control/package.xml new file: src/past_control/src/lane_follower_node.cpp new file: src/past_control/src/obstacle_detector_node.cpp new file: src/past_control/src/racing_orchestrator.cpp new file: src/planner/CMakeLists.txt new file: src/planner/config/planner.yaml new file: src/planner/launch/planner.launch.py new file: src/planner/package.xml new file: src/planner/src/planner_version.cpp modified: src/qr_detection/src/qr_dete_depth.cpp new file: src/racing_control/CMakeLists.txt new file: src/racing_control/include/racing_control/racing_control.hpp new file: src/racing_control/package.xml new file: src/racing_control/src/racing_control.cpp modified: src/vlm_detect/setup.py new file: src/vlm_detect/vlm_detect/__pycache__/__init__.cpython-310.pyc new file: src/vlm_detect/vlm_detect/__pycache__/tts_node.cpython-310.pyc new file: src/vlm_detect/vlm_detect/test_publisher.py new file: src/vlm_detect/vlm_detect/tts_node.py modified: src/vlm_detect/vlm_detect/vlm_node.py new file: tools/measure_turning_radius.py new file: tools/set_volume.py new file: tools/udp_to_cmdvel.py new file: tools/windows_keyboard_control.py new file: vlm_server.py new file: "\350\260\203\350\257\225\350\256\260\345\275\225.Assets/1.png" renamed: "\350\260\203\350\257\225\350\256\260\345\275\225.log" -> "\350\260\203\350\257\225\350\256\260\345\275\225.md"
2026-06-22 17:16:45 +08:00
parent a9bfeef59b
commit 779b32362a
51 changed files with 3974 additions and 398 deletions
--- a/src/vlm_detect/setup.py
+++ b/src/vlm_detect/setup.py
@@ -21,6 +21,8 @@ setup(
    entry_points={
      'console_scripts': [
        'vlm_node = vlm_detect.vlm_node:main',
+            'test_publisher = vlm_detect.test_publisher:main',
+            'tts_node = vlm_detect.tts_node:main',
      ],
    },
 )
--- a/src/vlm_detect/vlm_detect/pycache/init.cpython-310.pyc
+++ b/src/vlm_detect/vlm_detect/pycache/init.cpython-310.pyc
--- a/src/vlm_detect/vlm_detect/pycache/tts_node.cpython-310.pyc
+++ b/src/vlm_detect/vlm_detect/pycache/tts_node.cpython-310.pyc
--- a/src/vlm_detect/vlm_detect/test_publisher.py
+++ b/src/vlm_detect/vlm_detect/test_publisher.py
@@ -0,0 +1,88 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+测试发布者：发送图片和触发信号给 VLM 节点
+用法: ros2 run vlm_detect test_publisher --ros-args -p image_path:="/path/to/image.jpg"
+"""
+
+import rclpy
+from rclpy.node import Node
+from std_msgs.msg import Int32, String
+from sensor_msgs.msg import CompressedImage
+import cv2
+import time
+
+
+class TestPublisher(Node):
+    def __init__(self):
+        super().__init__("test_publisher")
+
+        # 声明参数
+        self.declare_parameter(
+            "image_path", "/home/sunrise/yiliao_ws/my_model/image.png"
+        )
+        self.declare_parameter("interval", 5.0)  # 每 N 秒触发一次
+
+        image_path = self.get_parameter("image_path").value
+        interval = self.get_parameter("interval").value
+
+        # 发布者
+        self.image_pub = self.create_publisher(CompressedImage, "/image_mjpeg", 10)
+        self.sign_pub = self.create_publisher(Int32, "/sign4return", 10)
+
+        # 订阅结果
+        self.result_sub = self.create_subscription(
+            String, "/vlm_result", self.result_callback, 10
+        )
+
+        # 加载图片
+        self.image_data = cv2.imread(image_path)
+        if self.image_data is None:
+            self.get_logger().error(f"无法读取图片: {image_path}")
+            raise FileNotFoundError(f"Image not found: {image_path}")
+
+        encode_params = [cv2.IMWRITE_JPEG_QUALITY, 90]
+        _, jpeg_data = cv2.imencode(".jpg", self.image_data, encode_params)
+        self.jpeg_bytes = jpeg_data.tobytes()
+        self.get_logger().info(f"已加载图片: {image_path}")
+
+        # 定时器
+        self.timer = self.create_timer(interval, self.timer_callback)
+        self.get_logger().info(f"每 {interval}s 发送一次图片和触发信号")
+
+    def timer_callback(self):
+        # 1. 发送压缩图像
+        img_msg = CompressedImage()
+        img_msg.header.stamp = self.get_clock().now().to_msg()
+        img_msg.format = "jpeg"
+        img_msg.data = self.jpeg_bytes
+        self.image_pub.publish(img_msg)
+        self.get_logger().info("已发送图片")
+
+        # 2. 等待一小段时间让订阅者收到图片
+        time.sleep(1.0)
+
+        # 3. 发送触发信号 (sign=9)
+        sign_msg = Int32()
+        sign_msg.data = 9
+        self.sign_pub.publish(sign_msg)
+        self.get_logger().info("已发送触发信号 (sign=9)，等待 VLM 结果...")
+
+    def result_callback(self, msg):
+        self.get_logger().info(f"🔍 VLM 识别结果: {msg.data}")
+
+
+def main(args=None):
+    rclpy.init(args=args)
+    node = TestPublisher()
+    try:
+        rclpy.spin(node)
+    except KeyboardInterrupt:
+        pass
+    finally:
+        node.destroy_node()
+        rclpy.shutdown()
+
+
+if __name__ == "__main__":
+    main()
--- a/src/vlm_detect/vlm_detect/tts_node.py
+++ b/src/vlm_detect/vlm_detect/tts_node.py
@@ -0,0 +1,42 @@
+#!/usr/bin/env python3
+import rclpy, subprocess, requests, os
+from rclpy.node import Node
+from std_msgs.msg import String
+
+VLM_HOST = "http://192.168.10.173:8000"
+# USB Audio Device (Card 1)
+AUDIO_SINK = "alsa_output.usb-C-Media_Electronics_Inc._USB_Audio_Device-00.analog-stereo"
+AUDIO_ENV = {**os.environ, "PULSE_SINK": AUDIO_SINK}
+
+class TTSNode(Node):
+    def __init__(self):
+        super().__init__("tts_node")
+        self.sub = self.create_subscription(String, "/vlm_result", self.callback, 10)
+        self.get_logger().info("TTS 播报节点已启动 (USB Audio Device, edge-tts 自然语音)")
+
+    def callback(self, msg):
+        text = msg.data
+        self.get_logger().info(f"播报: {text}")
+        try:
+            resp = requests.post(f"{VLM_HOST}/v1/tts",
+                json={"text": text, "voice": "zh-CN-XiaoxiaoNeural"}, timeout=60)
+            mp3 = "/tmp/tts_out.mp3"
+            with open(mp3, "wb") as f:
+                f.write(resp.content)
+            subprocess.Popen(["ffplay", "-nodisp", "-autoexit", mp3],
+                stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL,
+                env=AUDIO_ENV)
+        except Exception as e:
+            self.get_logger().error(f"TTS 失败,降级 espeak: {e}")
+            subprocess.Popen(["espeak-ng", "-v", "zh", "-s", "150", text],
+                env=AUDIO_ENV)
+
+def main(args=None):
+    rclpy.init(args=args)
+    node = TTSNode()
+    try: rclpy.spin(node)
+    except KeyboardInterrupt: pass
+    finally: node.destroy_node(); rclpy.shutdown()
+
+if __name__ == "__main__":
+    main()
--- a/src/vlm_detect/vlm_detect/vlm_node.py
+++ b/src/vlm_detect/vlm_detect/vlm_node.py
@@ -15,14 +15,14 @@ import numpy as np

 class VLMProcessor(Node):
    def __init__(self):
-        super().__init__('vlm_dtetct')
+        super().__init__('vlm_detect')
        
        # 初始化 OpenAI 客户端
        self.client = OpenAI(
-            base_url="http://192.168.1.103:8000/v1",
-            api_key="EMPTY"
+            base_url="http://192.168.10.173:8000/v1",  # 本地 API 地址
+            api_key="EMPTY",  # 不需要真实 API key
        )
-        
+     
        # ROS2 组件
        self.bridge = CvBridge()
        self.latest_image = None
@@ -31,7 +31,7 @@ class VLMProcessor(Node):
        # 订阅图像话题
        self.image_sub = self.create_subscription(
            CompressedImage,
-            '/image',
+            '/image_mjpeg',
            self.image_callback,
            10
        )
@@ -57,18 +57,20 @@ class VLMProcessor(Node):
        """保存最新的图像"""
        with self.image_lock:
            try:
+                # bridge = CvBridge()
                np_arr = np.frombuffer(msg.data, np.uint8)
                # 使用 OpenCV 解码图像
                cv_image = cv2.imdecode(np_arr, cv2.IMREAD_COLOR)
+                # cv_image =bridge.imgmsg_to_cv2(msg,desired_encoding='bgr8')
                self.latest_image = cv_image
                self.get_logger().debug("recive picture")
            except Exception as e:
-                self.get_logger().error("err picture")
+                self.get_logger().error(f"err picture: {e}")

    def sign_callback(self, msg):
        """处理触发信号"""
        if msg.data == 9:
-            self.get_logger().info("收到触发信号 (6)，开始处理图像...")
+            self.get_logger().info(f"收到触发信号 ({msg.data})，开始处理图像...")
            
            # 检查是否有可用图像
            with self.image_lock:
@@ -79,12 +81,12 @@ class VLMProcessor(Node):
                # 保存临时图像文件
                temp_path = "/tmp/vlm_temp_image.jpg"
                cv2.imwrite(temp_path, self.latest_image)
-                self.get_logger().info("已保存临时图像}")
+                self.get_logger().info(f"已保存临时图像: {temp_path}")
            
            # 处理图像
            try:
                description = self.process_image(temp_path)
-                self.get_logger().info(f"图像描述结果:")
+                self.get_logger().info(f"图像描述结果: {description}")
                
                # 发布结果
                result_msg = String()
@@ -94,7 +96,7 @@ class VLMProcessor(Node):
                # 清理临时文件
                os.remove(temp_path)
            except Exception as e:
-                self.get_logger().error(f"处理图像时出错:{e}")
+                self.get_logger().error(f"处理图像时出错: {e}")

    def process_image(self, image_path):
        """使用 VLM 模型处理图像"""
@@ -106,26 +108,26 @@ class VLMProcessor(Node):
        start_time = time.time()
        
        response = self.client.chat.completions.create(
-            model="/home/yyh/vllm_test/model/InternVL3-1B",
+            model="./OpenGVLab/InternVL3-1B/",
            messages=[
                {
                    "role": "user",
                    "content": [
-                        {"type": "text", "text": "描述图片中的动漫病人"},
+                        {"type": "text", "text": "描述图片中有一个病人的特征，字数控制在20字以内。"},
                        {
                            "type": "image_url",
                            "image_url": {
                                "url": f"data:image/jpeg;base64,{base64_image}"
-                            }
-                        }
+                            },
+                        },
                    ]
                }
            ],
-            max_tokens=100
+            max_tokens=100,
        )
        
        processing_time = time.time() - start_time
-        self.get_logger().info("VLM 处理耗时秒")
+        self.get_logger().info(f"VLM 处理耗时 {processing_time:.1f}s")
        
        return response.choices[0].message.content