Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 0 additions & 1 deletion src/2d-carla-tracking
Submodule 2d-carla-tracking deleted from 713a7a
48 changes: 38 additions & 10 deletions src/car_tracking/README.md
Original file line number Diff line number Diff line change
@@ -1,14 +1,42 @@
# Carla-LowQuality-Viewer
CARLA 0.9.14 低画质版专用可视化脚本,针对低配置设备优化,适配低画质 CARLA 的 API 差异,解决着色器崩溃、异常类找不到等问题,实现轻量化的摄像头画面可视化+车辆自动识别交通信号灯(红灯停、绿灯行)与自动行驶功能。
Carla 工具集
本目录包含两个独立的Carla脚本,可直接运行,互不依赖。

## 功能说明
### 核心功能
1. **2D低画质可视化**:针对低配置设备优化,适配CARLA 0.9.14低画质版的API差异,解决着色器崩溃、异常类找不到等问题。
2. **交通信号灯识别**:车辆可自动识别交通信号灯状态,实现红灯停、绿灯行的逻辑。
3. **车辆自动行驶**:结合信号灯识别功能,完成轻量化的自动行驶控制。
## 程序1:carla_2d_ground_truth.py
### 功能
将车辆3D包围盒投影为2D地面真值边界框并可视化,适用于目标检测算法验证与数据集采集。

### 环境要求
- Carla 版本:0.9.10~0.9.15
- Python 版本:3.7
- 依赖包:`opencv-python` `numpy` `carla`(版本需与Carla模拟器匹配)

### 运行命令
```bash
# 第一步:启动对应版本的Carla模拟器
# 第二步:执行脚本
python car_2d_ground_truth.py
```

### 操作说明
- 可视化窗口显示相机画面与车辆2D边界框
- 按 **q** 键退出程序

## 程序2:carla_low_quality_viewer.py
### 功能
Carla 0.9.14低画质版专用脚本,实现车辆自动行驶+提前避让(车辆/障碍物/红绿灯),修复低配置设备进程卡顿问题,降低资源占用。

### 环境要求
- Carla 版本:**仅限0.9.14低画质版**
- Python 版本:3.7
- 依赖包:`opencv-python` `numpy` `carla==0.9.14`(可选`psutil`用于进程检测)

## 快速开始
### 运行命令
在项目目录下执行以下命令,启动可视化与自动行驶功能:
```bash
python main.py
# 第一步:启动Carla 0.9.14低画质版模拟器
# 第二步:执行脚本
python main.py
```

### 操作说明
- 可视化窗口显示摄像头画面,终端实时输出车速与避让提示
- 按 **q** 键退出程序
265 changes: 265 additions & 0 deletions src/car_tracking/car1.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,265 @@
import carla
import queue
import random
import cv2
import numpy as np

# 注意:移除了所有deep_sort相关导入
# 保留必要的工具函数导入(如果utils模块不存在,我会替换为内置实现)
from utils.projection import build_projection_matrix, get_image_point, get_2d_box_from_3d_edges, point_in_canvas
from utils.world import clear_npc, clear_static_vehicle

# ---------------------------
# 全局配置(可根据需要调整)
# ---------------------------
MAP_NAME = "Town07" # 指定要加载的地图,可替换为Town01-Town12
NPC_VEHICLE_NUM = 50 # NPC车辆数量
CAMERA_RESOLUTION = (640, 640) # 相机分辨率
SYNC_DELTA_SECONDS = 0.05 # 同步模式的固定时间步长
VEHICLE_DISTANCE_THRESHOLD = 50 # 检测车辆的距离阈值(米)
SPEED_DISPLAY_UNIT = "km/h" # 速度单位:m/s 或 km/h

# ---------------------------
# 替代draw_bounding_boxes的内置函数(如果utils.box_utils不存在)
# 如果你有utils.box_utils,可注释这部分,保留原导入
# ---------------------------
def draw_bounding_boxes(image, bboxes, labels, class_names, ids):
"""
绘制包围盒和ID标签
:param image: 输入图像(BGR格式)
:param bboxes: 包围盒列表,格式为[[x1,y1,x2,y2], ...]
:param labels: 类别标签列表
:param class_names: 类别名称列表
:param ids: 跟踪ID列表
:return: 绘制后的图像
"""
for i, (bbox, id) in enumerate(zip(bboxes, ids)):
x1, y1, x2, y2 = bbox.astype(int)
# 绘制矩形框
cv2.rectangle(image, (x1, y1), (x2, y2), (0, 255, 0), 2)
# 绘制标签(类别 + ID)
label = f"{class_names[labels[i]]} | ID: {id}"
cv2.putText(image, label, (x1, y1 - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
return image

def camera_callback(image, rgb_image_queue):
"""相机回调函数,将图像数据存入队列"""
rgb_image_queue.put(np.reshape(np.copy(image.raw_data),
(image.height, image.width, 4)))

def calculate_vehicle_speed(velocity, unit="km/h"):
"""计算车辆速度,转换为指定单位"""
speed_m_s = np.linalg.norm([velocity.x, velocity.y, velocity.z])
if unit == "km/h":
return speed_m_s * 3.6 # 1 m/s = 3.6 km/h
return speed_m_s

def main():
# ---------------------------
# 1. 初始化Carla客户端,加快地图加载
# ---------------------------
try:
client = carla.Client('localhost', 2000)
client.set_timeout(30.0) # 增加超时时间,避免地图加载时超时

# 方案3:查看所有可用地图
print("Carla支持的地图列表:")
available_maps = client.get_available_maps()
print(available_maps)

# 方案2:使用当前已加载的地图,避免map not found
world = client.get_world()
# 如果想指定地图,可从available_maps中选一个,比如:
# world = client.load_world("Town01")

spawn_points = world.get_map().get_spawn_points()
if not spawn_points:
print("警告:未获取到地图生成点,程序将退出")
return
# 后续代码不变...

# ---------------------------
# 2. 设置同步模式,优化仿真效率
# ---------------------------
settings = world.get_settings()
settings.synchronous_mode = True
settings.fixed_delta_seconds = SYNC_DELTA_SECONDS
world.apply_settings(settings)

# ---------------------------
# 3. 生成Ego车辆
# ---------------------------
bp_lib = world.get_blueprint_library()
vehicle_bp = bp_lib.find('vehicle.lincoln.mkz_2020')
ego_vehicle = world.try_spawn_actor(vehicle_bp, random.choice(spawn_points))
if not ego_vehicle:
print("警告:未能生成Ego车辆,程序将退出")
return

# ---------------------------
# 4. 生成相机传感器
# ---------------------------
camera_bp = bp_lib.find('sensor.camera.rgb')
camera_bp.set_attribute('image_size_x', str(CAMERA_RESOLUTION[0]))
camera_bp.set_attribute('image_size_y', str(CAMERA_RESOLUTION[1]))
camera_init_trans = carla.Transform(carla.Location(x=1, z=2))
camera = world.spawn_actor(camera_bp, camera_init_trans, attach_to=ego_vehicle)

# 图像队列(设置最大长度,避免内存溢出)
image_queue = queue.Queue(maxsize=1)
camera.listen(lambda image: camera_callback(image, image_queue))

# ---------------------------
# 5. 清理现有NPC和静态车辆
# ---------------------------
clear_npc(world)
clear_static_vehicle(world)

# ---------------------------
# 6. 预计算相机相关参数
# ---------------------------
edges = [[0, 1], [1, 3], [3, 2], [2, 0], [0, 4], [4, 5],
[5, 1], [5, 7], [7, 6], [6, 4], [6, 2], [7, 3]]
image_w, image_h = CAMERA_RESOLUTION
fov = camera_bp.get_attribute("fov").as_float()
K = build_projection_matrix(image_w, image_h, fov)
K_b = build_projection_matrix(image_w, image_h, fov, is_behind_camera=True)

# ---------------------------
# 7. 生成NPC车辆
# ---------------------------
car_bp = [bp for bp in bp_lib.filter('vehicle') if int(bp.get_attribute('number_of_wheels')) == 4]
if car_bp:
for i in range(NPC_VEHICLE_NUM):
npc = world.try_spawn_actor(random.choice(car_bp), random.choice(spawn_points))
if npc:
npc.set_autopilot(True)

ego_vehicle.set_autopilot(True)

# ---------------------------
# 8. 主循环(核心:用Actor ID实现跟踪)
# ---------------------------
while True:
world.tick()

# 移动spectator到车辆上方
spectator = world.get_spectator()
transform = carla.Transform(ego_vehicle.get_transform().transform(
carla.Location(x=-4, z=50)), carla.Rotation(yaw=-180, pitch=-90))
spectator.set_transform(transform)

# 获取相机图像
if not image_queue.empty():
image = image_queue.get()
else:
continue

# 图像预处理(RGBA转BGR,减少一次颜色转换)
image = cv2.cvtColor(image, cv2.COLOR_RGBA2BGR)

# 更新相机的世界到相机矩阵
world_2_camera = np.array(camera.get_transform().get_inverse_matrix())

# 存储:包围盒、车辆ID、速度
bboxes = []
vehicle_ids = []
vehicle_speeds = []

# 遍历所有车辆Actor
vehicles = world.get_actors().filter('*vehicle*')
for npc in vehicles:
if npc.id == ego_vehicle.id:
# 输出Ego车辆速度
ego_speed = calculate_vehicle_speed(npc.get_velocity(), SPEED_DISPLAY_UNIT)
print(f"Ego车辆速度:{ego_speed:.1f} {SPEED_DISPLAY_UNIT}")
continue

# 过滤距离过远的车辆
dist = npc.get_transform().location.distance(ego_vehicle.get_transform().location)
if dist > VEHICLE_DISTANCE_THRESHOLD:
continue

# 只检测车辆前方的目标
forward_vec = ego_vehicle.get_transform().get_forward_vector()
ray = npc.get_transform().location - ego_vehicle.get_transform().location
if forward_vec.dot(ray) <= 0:
continue

# 计算NPC车辆速度
npc_speed = calculate_vehicle_speed(npc.get_velocity(), SPEED_DISPLAY_UNIT)

# 计算3D包围盒的2D投影
bb = npc.bounding_box
verts = [v for v in bb.get_world_vertices(npc.get_transform())]
points_2d = []

for vert in verts:
ray0 = vert - camera.get_transform().location
cam_forward_vec = camera.get_transform().get_forward_vector()
if cam_forward_vec.dot(ray0) > 0:
p = get_image_point(vert, K, world_2_camera)
else:
p = get_image_point(vert, K_b, world_2_camera)
points_2d.append(p)

# 获取2D框
x_min, x_max, y_min, y_max = get_2d_box_from_3d_edges(points_2d, edges, image_h, image_w)

# 过滤过小的框和超出画布的框
box_area = (y_max - y_min) * (x_max - x_min)
box_width = x_max - x_min
if box_area > 100 and box_width > 20 and point_in_canvas((x_min, y_min), image_h, image_w) and point_in_canvas((x_max, y_max), image_h, image_w):
bboxes.append(np.array([x_min, y_min, x_max, y_max]))
vehicle_ids.append(npc.id) # 用Actor ID作为跟踪ID
vehicle_speeds.append(npc_speed)

# ---------------------------
# 绘制包围盒、ID和速度(核心替代deep_sort的部分)
# ---------------------------
if len(bboxes) > 0:
# 绘制基础包围盒和ID(类别固定为car,对应COCO的2号类别)
labels = np.array([2] * len(bboxes))
COCO_CLASS_NAMES = {2: 'car'} # 简化的类别映射,替代原导入
image = draw_bounding_boxes(image, bboxes, labels, COCO_CLASS_NAMES, vehicle_ids)

# 绘制速度标签
for i, (bbox, speed) in enumerate(zip(bboxes, vehicle_speeds)):
x1, y1, x2, y2 = bbox.astype(int)
speed_label = f"Speed: {speed:.1f} {SPEED_DISPLAY_UNIT}"
# 在框下方绘制速度
cv2.putText(image, speed_label, (x1, y2 + 20), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 0, 0), 2)

# 显示图像
cv2.imshow('2D Tracking with Actor ID (no deep_sort)', image)

# 按q退出
if cv2.waitKey(1) & 0xFF == ord('q'):
break

except KeyboardInterrupt:
print("\n程序被用户中断")
except Exception as e:
print(f"程序出错:{e}")
finally:
# ---------------------------
# 资源清理
# ---------------------------
print("正在清理资源...")
if 'world' in locals():
settings = world.get_settings()
settings.synchronous_mode = False
world.apply_settings(settings)
if 'camera' in locals():
camera.destroy()
if 'ego_vehicle' in locals():
ego_vehicle.destroy()
# 清理所有NPC车辆
if 'world' in locals():
for vehicle in world.get_actors().filter('*vehicle*'):
vehicle.destroy()
cv2.destroyAllWindows()
print("资源清理完成")

if __name__ == "__main__":
main()
Loading
Loading