carla学习笔记(五)_carla.colorconverter

今天主要学习了example里面的tutorial.py的源码，越发觉得这才应该是最开始学习的代码，大部分代码块都有注释，理解起来十分容易。

#!/usr/bin/env python
 
# Copyright (c) 2019 Computer Vision Center (CVC) at the Universitat Autonoma de
# Barcelona (UAB).
#
# This work is licensed under the terms of the MIT license.
# For a copy, see <https://opensource.org/licenses/MIT>.
 
import glob
import os
import sys
 
try:
    sys.path.append(glob.glob('../carla/dist/carla-*%d.%d-%s.egg' % (
        sys.version_info.major,
        sys.version_info.minor,
        'win-amd64' if os.name == 'nt' else 'linux-x86_64'))[0])
except IndexError:
    pass
 
import carla
 
import random
import time
 
 
def main():
    actor_list = []
 
    # In this tutorial script, we are going to add a vehicle to the simulation
    # and let it drive in autopilot. We will also create a camera attached to
    # that vehicle, and save all the images generated by the camera to disk.
 
    try:
        # First of all, we need to create the client that will send the requests
        # to the simulator. Here we'll assume the simulator is accepting
        # requests in the localhost at port 2000.
        client = carla.Client('localhost', 2000)
        client.set_timeout(2.0)
 
        # Once we have a client we can retrieve the world that is currently
        # running.
        world = client.get_world()
 
        # The world contains the list blueprints that we can use for adding new
        # actors into the simulation.
        blueprint_library = world.get_blueprint_library()
 
        # Now let's filter all the blueprints of type 'vehicle' and choose one
        # at random.
        bp = random.choice(blueprint_library.filter('vehicle'))
 
        # A blueprint contains the list of attributes that define a vehicle's
        # instance, we can read them and modify some of them. For instance,
        # let's randomize its color.
        if bp.has_attribute('color'):
            color = random.choice(bp.get_attribute('color').recommended_values)
            bp.set_attribute('color', color)
 
        # Now we need to give an initial transform to the vehicle. We choose a
        # random transform from the list of recommended spawn points of the map.
        transform = random.choice(world.get_map().get_spawn_points())
 
        # So let's tell the world to spawn the vehicle.
        vehicle = world.spawn_actor(bp, transform)
 
        # It is important to note that the actors we create won't be destroyed
        # unless we call their "destroy" function. If we fail to call "destroy"
        # they will stay in the simulation even after we quit the Python script.
        # For that reason, we are storing all the actors we create so we can
        # destroy them afterwards.
        # 这里是为了帮助销毁vehicles的时候方便，特地设置了一个list。
        actor_list.append(vehicle)
        print('created %s' % vehicle.type_id)
 
        # Let's put the vehicle to drive around.
        vehicle.set_autopilot(True)
 
        # Let's add now a "depth" camera attached to the vehicle. Note that the
        # transform we give here is now relative to the vehicle.
        camera_bp = blueprint_library.find('sensor.camera.depth')
        cam_bp = blueprint_library.find('sensor.camera.rgb')
        
        camera_transform = carla.Transform(carla.Location(z=20))
        cam01_transform = carla.Transform(carla.Location(x=10, y=10, z=20))
 
        # camera = world.spawn_actor(camera_bp, camera_transform, attach_to=vehicle)
        camera = world.spawn_actor(camera_bp, camera_transform)
        cam01 = world.spawn_actor(cam_bp, cam01_transform)
        actor_list.append(camera)
        actor_list.append(cam01)
 
        print('created %s' % camera.type_id)
        print('created %s' % cam01.type_id)
 
        # Now we register the function that will be called each time the sensor
        # receives an image. In this example we are saving the image to disk
        # converting the pixels to gray-scale.
        # cc = carla.ColorConverter.LogarithmicDepth
        # camera.listen(lambda image: image.save_to_disk('_out/%06d.png' % image.frame, cc))
        cam01.listen(lambda image: image.save_to_disk('_out/%06d.png' % image.frame))
        # camera.listen(lambda image: image.save_to_disk('_out/%06d.png' % image.frame))
 
 
        # Oh wait, I don't like the location we gave to the vehicle, I'm going
        # to move it a bit forward.
        location = vehicle.get_location()
        location.x += 40
        vehicle.set_location(location)
        print('moved vehicle to %s' % location)
 
        # But the city now is probably quite empty, let's add a few more
        # vehicles.
        # 创造一列车10辆
        transform.location += carla.Location(x=40, y=-3.2)
        transform.rotation.yaw = -180.0
        # for _ in range(0, 10):
        #     transform.location.x += 8.0
 
        #     bp = random.choice(blueprint_library.filter('vehicle'))
 
        #     # This time we are using try_spawn_actor. If the spot is already
        #     # occupied by another object, the function will return None.
        #     npc = world.try_spawn_actor(bp, transform)
        #     if npc is not None:
        #         actor_list.append(npc)
        #         npc.set_autopilot(True)
        #         print('created %s' % npc.type_id)
 
        time.sleep(500)
 
    finally:
 
        print('destroying actors')
        camera.destroy()
        client.apply_batch([carla.command.DestroyActor(x) for x in actor_list])
        print('done.')
 
 
if __name__ == '__main__':
 
    main()

然后我看到它有一部分里面有存储sensor采集到的数据的部分，我就想起了昨天的问题，准备借鉴这部分的内容在验证一下：

"""
学习tutorial.py时，看到了类似功能的代码，然后检验了一下我昨天的设想。
主要内容是：
生成一个camera.depth、一个camera.rgb、一个lidar然后分别存储进actor_list[]，最后保存传感器采集
的数据，然后返回matlab显示。采集的数量目前来看是通过time.sleep()函数来控制的。
"""
 
import glob
import os
import sys
 
try:
    sys.path.append(glob.glob('../carla/dist/carla-*%d.%d-%s.egg' % (
        sys.version_info.major,
        sys.version_info.minor,
        'win-amd64' if os.name == 'nt' else 'linux-x86_64'))[0])
except IndexError:
    pass
 
import carla
import random
import time
 
 
def main():
    actor_list = []
 
    try:
        client = carla.Client('localhost', 2000)
        client.set_timeout(2.0)
 
        world = client.get_world()
 
        blueprint_library = world.get_blueprint_library()
 
        bp = random.choice(blueprint_library.filter('vehicle'))
 
        if bp.has_attribute('color'):
            color = random.choice(bp.get_attribute('color').recommended_values)
            bp.set_attribute('color', color)
 
        transform = random.choice(world.get_map().get_spawn_points())
 
        # 在蓝图库中找到三个传感器
        camera_bp = blueprint_library.find('sensor.camera.depth')
        cam_bp = blueprint_library.find('sensor.camera.rgb')
        lidar_bp = blueprint_library.find('sensor.lidar.ray_cast')
        # 设置三个传感器的生成的位置
        camera_transform = carla.Transform(carla.Location(z=20))
        cam01_transform = carla.Transform(carla.Location(x=10, y=10, z=20))
        lidar01_transform = carla.Transform(carla.Location(x=20, y=10, z=1))
        # 这个location的位置选的不太好，居然一个点都没有采到，不知道是什么问题，现在验证一下。利用draw_string函数：
        # world.debug.draw_string(lidar01_transform.location, 'O', draw_shadow=False,
        #                 color=carla.Color(r=0, g=255, b=0), life_time=500,
        #                 persistent_lines=True)
        # 发现了是z设置的问题，一开始设置的太高了，设置了20
 
        # 生成三个传感器，独立于vehicle
        camera = world.spawn_actor(camera_bp, camera_transform)
        cam01 = world.spawn_actor(cam_bp, cam01_transform)
        lidar01 = world.spawn_actor(lidar_bp, lidar01_transform)
 
        #在actor_list表中添加三个传感器 
        actor_list.append(camera)
        actor_list.append(cam01)
        actor_list.append(lidar01)
 
        # 终端输出三个传感器的名字id
        print('created %s' % camera.type_id)
        print('created %s' % cam01.type_id)
        print('created %s' % lidar01.type_id)
 
        # Now we register the function that will be called each time the sensor
        # receives an image. In this example we are saving the image to disk
        # converting the pixels to gray-scale.
        # cc = carla.ColorConverter.LogarithmicDepth
        # camera.listen(lambda image: image.save_to_disk('_out/%06d.png' % image.frame, cc))
        # cam01.listen(lambda image: image.save_to_disk('/media/hhh/75c0c2a9-f565-4a05-b2a5-5599a918e2f0/hhh/carlaLearning/PythonAPI/learning_document/%06d.png' % image.frame))
        # 解决了昨天的疑问，应该是昨天存储路径的问题，就是os.path的问题，我的设想从一开始就没有问题。
        
        # camera.listen(lambda image: image.save_to_disk('_out/%06d.png' % image.frame))
        # lidar01.listen(lambda image: image.save_to_disk('/media/hhh/75c0c2a9-f565-4a05-b2a5-5599a918e2f0/hhh/carlaLearning/PythonAPI/learning_document/%06d.ply' % image.frame))
 
        # But the city now is probably quite empty, let's add a few more
        # vehicles.
        # 创造一列车,10辆
        # 但我发现每次的都是随机生成的，猜测是carla随机生成
        # 后来仔细看来代码，transform.location += carla.Location(x=40, y=-3.2),这里的transform应该
        # 是用了前面的random选择，这里的carla.Location应该是x=40,y=-3.2,z=0然后加到前面的transform上面去
        transform.location += carla.Location(x=40, y=-3.2)
        transform.rotation.yaw = -180.0
        print(transform)
        for _ in range(0, 10):
            transform.location.x += 8.0
            print(transform)
            bp = random.choice(blueprint_library.filter('vehicle'))
        
            # This time we are using try_spawn_actor. If the spot is already
            # occupied by another object, the function will return None.
            # 这个函数挺有意思的，大概封装了碰撞检测函数？
            npc = world.try_spawn_actor(bp, transform)
            if npc is not None:
                actor_list.append(npc)
                npc.set_autopilot(True)
                print('created %s' % npc.type_id)
 
        #这个sleep设置了try部分运行的时间。 
        time.sleep(500)
 
    finally:
 
        print('destroying actors')
        camera.destroy()
        # 利用指令来删除actor_list里面的vehicles
        client.apply_batch([carla.command.DestroyActor(x) for x in actor_list])
        print('done.')
 
 
if __name__ == '__main__':
 
    main()

最后显示是可以采集的，也可以正常存储。说明项目要求的路侧放置lidar基本可以实现。下周周报有东西可以写了。准备这段时间尝试在路侧放置一个lidar，然后控制一辆车从路口经过，然后分析lidar采集的数据，看能不能清晰的实现。