basic

Link to the world model

import sys
import time
import glob
import os
import random
import numpy as np
import math
from collections import deque
from threading import Thread

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 cv2
client = carla.Client("localhost",2000)
client.timeout = 2.0
world = client.get_world()

Spawn_point

# Get the blueprint library and the spawn points for the map
bp_lib = world.get_blueprint_library() 
spawn_points = world.get_map().get_spawn_points() # a list
print(spawn_points[1]) 
"""
Transform(Location(x=-67.254570, y=27.963758, z=0.600000), Rotation(pitch=0.000000, yaw=0.159198, roll=0.000000))
"""

Select car

vehicle_bp = bp_lib.find('vehicle.lincoln.mkz_2020') 

# Try spawning the vehicle at a randomly chosen spawn point
vehicle = world.try_spawn_actor(vehicle_bp, random.choice(spawn_points))

"""
# Get all blueprint objects
blueprint_objects = blueprint_library.filter('vehicle') ## sensor, 

# Print the names of all blueprint objects
for blueprint_object in blueprint_objects:
    print(blueprint_object.id)
sensor.other.gnss
sensor.lidar.ray_cast
sensor.camera.semantic_segmentation
sensor.other.radar
sensor.other.lane_invasion
sensor.camera.instance_segmentation
sensor.camera.rgb
sensor.other.rss
sensor.camera.optical_flow
sensor.lidar.ray_cast_semantic
sensor.other.obstacle
sensor.other.imu
sensor.camera.depth
sensor.camera.dvs
sensor.other.collision
"""

change view

spectator = world.get_spectator() 


transform = carla.Transform(vehicle.get_transform().transform(carla.Location(x=-4,z=2.5)),vehicle.get_transform().rotation) 
spectator.set_transform(transform) 

Add some cars

for i in range(300): 
    vehicle_bp = random.choice(bp_lib.filter('vehicle')) 
    npc = world.try_spawn_actor(vehicle_bp, random.choice(spawn_points)) 
for v in world.get_actors().filter('*vehicle*'): 
    v.set_autopilot(True) 
"""
Actor(id=164, type=vehicle.citroen.c3)
Actor(id=163, type=vehicle.ford.crown)
Actor(id=162, type=vehicle.nissan.micra)
Actor(id=161, type=vehicle.tesla.cybertruck)
Actor(id=160, type=vehicle.mercedes.sprinter)
Actor(id=159, type=vehicle.carlamotors.firetruck)
Actor(id=158, type=vehicle.seat.leon)
Actor(id=157, type=vehicle.kawasaki.ninja)
Actor(id=156, type=vehicle.ford.ambulance)
Actor(id=155, type=vehicle.toyota.prius)
Actor(id=154, type=vehicle.dodge.charger_police_2020)
Actor(id=153, type=vehicle.audi.etron)
Actor(id=152, type=vehicle.toyota.prius)
Actor(id=151, type=vehicle.dodge.charger_2020)
Actor(id=150, type=vehicle.vespa.zx125)
Actor(id=149, type=vehicle.citroen.c3)
Actor(id=148, type=vehicle.lincoln.mkz_2017)
Actor(id=147, type=vehicle.harley-davidson.low_rider)
Actor(id=146, type=vehicle.harley-davidson.low_rider)
Actor(id=145, type=vehicle.carlamotors.carlacola)
Actor(id=144, type=vehicle.carlamotors.firetruck)
"""

Destory all actors in carla

actors = world.get_actors()

# For each actor in the list, call the destroy() method.
for actor in actors:
    actor.destroy()

Scene

Change town

python3 config.py --map Town01

py config.py --list
weather presets:

    ClearNight, ClearNoon, ClearSunset, CloudyNight, CloudyNoon,
    CloudySunset, Default, HardRainNight, HardRainNoon,
    HardRainSunset, MidRainSunset, MidRainyNight, MidRainyNoon,
    SoftRainNight, SoftRainNoon, SoftRainSunset, WetCloudyNight,
    WetCloudyNoon, WetCloudySunset, WetNight, WetNoon, WetSunset.

Retrieve simulation data

# carmera
camera.listen(lambda image: image.save_to_disk('out/%06d.png' % image.frame))

# video
import cv2
import numpy as np
# 注意帧速率和aspect ratio 必须和sensor 完全一致
video_writer = cv2.VideoWriter('out/video.avi', cv2.VideoWriter_fourcc(*"MJPG"), 30, (1920, 1080))
def save_image(image):
    # Convert the CARLA image to a numpy array
    transform = carla.Transform(vehicle.get_transform().transform(carla.Location(x=-4,z=2.5)),vehicle.get_transform().rotation) 
    spectator.set_transform(transform) 
    np_image = np.array(image.raw_data)

    # Reshape the array to match the image height, width, and number of channels
    reshaped_image = np_image.reshape((image.height, image.width, 4))
    # Convert RGBA to BGR (OpenCV uses BGR color format)
    bgr_image = reshaped_image[:, :, :3]
    # Write the BGR image to the video file
    video_writer.write(bgr_image)

# Assuming you have a camera sensor object named 'camera'
camera.listen(lambda image: save_image(image))

Traffic manager

use of 同步模式

Map

download roadrunner

wget https://github.com/roadrunner-server/roadrunner/releases/download/v2023.1.3/roadrunner-2023.1.3-linux-amd64.deb
sudo dpkg -i roadrunner-2023.1.3-linux-amd64.deb