Implementation:CARLA simulator Carla SensorData API Spec

Knowledge Sources	CARLA
Domains	Simulation, Sensors, API Specification
Last Updated	2026-02-15 05:00 GMT

Overview

YML specification file that defines sensor data classes in the CARLA Python API, including SensorData, Image, ColorConverter, CityObjectLabel, and various measurement types.

Description

This YML file specifies the data classes generated by CARLA sensors. SensorData is the base class with frame (int), timestamp (float, seconds), and transform properties. ColorConverter provides image conversion patterns (CityScapesPalette, Depth, LogarithmicDepth, Raw). CityObjectLabel enumerates semantic tags for bounding box filtering (Buildings, Vehicles, Pedestrians, etc.). Image (inheriting from SensorData) represents 32-bit BGRA camera output with fov, height, width, and raw_data properties, plus convert() and save_to_disk() methods. The file also covers CollisionEvent, GnssMeasurement, IMUMeasurement, LaneInvasionEvent, LidarMeasurement, ObstacleDetectionEvent, RadarMeasurement, and SemanticLidarMeasurement data types.

Usage

Use these classes to process data from CARLA sensors in callback functions. Each sensor type produces a specific data subclass that should be handled in the sensor's listen() callback.

Code Reference

Source Location

Repository: CARLA
File: PythonAPI/docs/sensor_data.yml

Signature

- module_name: carla
  classes:
  - class_name: SensorData
    instance_variables:
    - var_name: frame        # type: int
    - var_name: timestamp    # type: float (seconds)
    - var_name: transform    # type: carla.Transform

  - class_name: ColorConverter
    instance_variables:
    - var_name: CityScapesPalette
    - var_name: Depth
    - var_name: LogarithmicDepth
    - var_name: Raw

  - class_name: CityObjectLabel
    instance_variables:
    - var_name: Buildings
    - var_name: Pedestrians
    - var_name: Vehicles
    - var_name: TrafficSigns
    - var_name: TrafficLight
    - var_name: Any

  - class_name: Image
    parent: carla.SensorData
    instance_variables:
    - var_name: fov          # type: float (degrees)
    - var_name: height       # type: int
    - var_name: width        # type: int
    - var_name: raw_data     # type: bytes
    methods:
    - def_name: convert
    - def_name: save_to_disk

Import

import carla
# Sensor data is received via callbacks:
camera.listen(lambda image: image.save_to_disk('output/%06d.png' % image.frame))

I/O Contract

Class	Source Sensor	Key Properties
Image	RGB/Depth/Semantic Camera	fov, height, width, raw_data
CollisionEvent	Collision Detector	other_actor, normal_impulse
GnssMeasurement	GNSS Sensor	latitude, longitude, altitude
IMUMeasurement	IMU Sensor	accelerometer, gyroscope, compass
LaneInvasionEvent	Lane Invasion Detector	crossed_lane_markings
LidarMeasurement	LIDAR Sensor	channels, raw_data
RadarMeasurement	Radar Sensor	raw_data (velocity, azimuth, altitude, depth)
SemanticLidarMeasurement	Semantic LIDAR	channels, raw_data with semantic info

Usage Examples

import carla
import numpy as np

client = carla.Client('localhost', 2000)
world = client.get_world()

# Spawn an RGB camera
bp = world.get_blueprint_library().find('sensor.camera.rgb')
bp.set_attribute('image_size_x', '800')
bp.set_attribute('image_size_y', '600')
transform = carla.Transform(carla.Location(x=1.5, z=2.4))
vehicle = world.get_actors().filter('vehicle.*')[0]
camera = world.spawn_actor(bp, transform, attach_to=vehicle)

# Process image data
def process_image(image):
    array = np.frombuffer(image.raw_data, dtype=np.uint8)
    array = array.reshape((image.height, image.width, 4))
    image.save_to_disk('output/%06d.png' % image.frame)

camera.listen(process_image)

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment