Implementation:CARLA simulator Carla TrafficManager Configuration

Knowledge Sources	CARLA Simulator CARLA Documentation
Domains	Traffic Behavior Modeling, Simulation Configuration
Last Updated	2026-02-15 00:00 GMT

Overview

Concrete tool for configuring the Traffic Manager's global and per-vehicle behavior parameters, provided by the CARLA simulator.

Description

The TrafficManager object exposes multiple configuration methods that control how NPC vehicles drive. These methods are grouped into:

• Global settings -- Apply to all vehicles managed by this Traffic Manager instance
• Per-vehicle settings -- Override global defaults for specific vehicles
• Mode settings -- Control synchronous execution, hybrid physics, and determinism

The key configuration methods covered here are:

• set_global_distance_to_leading_vehicle(distance) -- Sets the minimum following distance (in meters) for all managed vehicles.
• global_percentage_speed_difference(percentage) -- Adjusts the target speed of all managed vehicles relative to the speed limit. Positive values slow down, negative values speed up.
• set_synchronous_mode(mode) -- Enables or disables synchronous execution, ensuring the TM processes updates in lockstep with world ticks.
• set_hybrid_physics_mode(enabled) -- Enables or disables hybrid physics, where vehicles far from the hero vehicle use simplified kinematic movement.
• set_hybrid_physics_radius(radius) -- Sets the distance (in meters) from the hero vehicle within which full physics is applied.
• set_random_device_seed(seed) -- Sets the random seed for deterministic traffic behavior.

Usage

Call these methods on the TrafficManager object after obtaining it via Client.get_trafficmanager(). Configuration should ideally be done before spawning NPC vehicles, though most parameters can be changed at runtime.

Code Reference

Source Location

• Repository: CARLA
• File: LibCarla/source/carla/trafficmanager/TrafficManager.h
• Lines: L150-360
• Python Binding: PythonAPI/carla/src/TrafficManager.cpp, L76-113

Signature

TrafficManager.set_global_distance_to_leading_vehicle(distance: float) -> None
TrafficManager.global_percentage_speed_difference(percentage: float) -> None
TrafficManager.set_synchronous_mode(mode: bool) -> None
TrafficManager.set_hybrid_physics_mode(enabled: bool) -> None
TrafficManager.set_hybrid_physics_radius(radius: float) -> None
TrafficManager.set_random_device_seed(seed: int) -> None

Import

import carla

I/O Contract

Inputs

Name	Type	Required	Description
distance	float	Yes	Minimum following distance in meters. Default is 2.5m. Typical range: 1.0 to 10.0.
percentage	float	Yes	Speed difference as a percentage of the speed limit. Positive values reduce speed (e.g., 30.0 = 30% slower). Negative values increase speed (e.g., -10.0 = 10% faster). Range: -100.0 to 100.0.
mode	bool	Yes	True to enable synchronous mode, False to disable. Should match the world's synchronous mode setting.
enabled	bool	Yes	True to enable hybrid physics mode, False for full physics on all vehicles.
radius	float	Yes	Distance in meters from the hero vehicle within which full physics simulation is applied. Default is 50.0m. Vehicles outside this radius use kinematic movement.
seed	int	Yes	Random seed for deterministic behavior. Any integer value. Use the same seed across runs for reproducibility.

Outputs

Name	Type	Description
return (all methods)	None	All configuration methods return None. Changes take effect on the next simulation tick.

Usage Examples

Basic Example

import carla

client = carla.Client('localhost', 2000)
client.set_timeout(10.0)
world = client.get_world()

# Initialize Traffic Manager
traffic_manager = client.get_trafficmanager(8000)

# Configure global behavior
traffic_manager.set_global_distance_to_leading_vehicle(3.0)  # 3 meter gap
traffic_manager.global_percentage_speed_difference(20.0)      # 20% below speed limit

Full Deterministic Configuration

import carla

client = carla.Client('localhost', 2000)
client.set_timeout(10.0)
world = client.get_world()

# Enable synchronous mode on the world
settings = world.get_settings()
settings.synchronous_mode = True
settings.fixed_delta_seconds = 0.05  # 20 Hz fixed time step
world.apply_settings(settings)

# Configure Traffic Manager for deterministic simulation
traffic_manager = client.get_trafficmanager(8000)
traffic_manager.set_synchronous_mode(True)
traffic_manager.set_random_device_seed(42)

# Set driving behavior
traffic_manager.set_global_distance_to_leading_vehicle(2.5)
traffic_manager.global_percentage_speed_difference(10.0)

print("Deterministic traffic configuration complete")

High-Performance Configuration with Hybrid Physics

import carla

client = carla.Client('localhost', 2000)
client.set_timeout(10.0)
world = client.get_world()

# Synchronous mode
settings = world.get_settings()
settings.synchronous_mode = True
settings.fixed_delta_seconds = 0.05
world.apply_settings(settings)

traffic_manager = client.get_trafficmanager(8000)
traffic_manager.set_synchronous_mode(True)
traffic_manager.set_random_device_seed(0)

# Enable hybrid physics for large-scale traffic
# Only vehicles within 70m of the hero vehicle get full physics
traffic_manager.set_hybrid_physics_mode(True)
traffic_manager.set_hybrid_physics_radius(70.0)

# Moderate driving behavior
traffic_manager.set_global_distance_to_leading_vehicle(2.0)
traffic_manager.global_percentage_speed_difference(15.0)

print("Hybrid physics enabled: full physics within 70m of hero vehicle")

Per-Vehicle Configuration Overrides

import carla
import random

client = carla.Client('localhost', 2000)
client.set_timeout(10.0)
world = client.get_world()

traffic_manager = client.get_trafficmanager(8000)
traffic_manager.set_synchronous_mode(True)
tm_port = traffic_manager.get_port()

# Set conservative global defaults
traffic_manager.set_global_distance_to_leading_vehicle(3.5)
traffic_manager.global_percentage_speed_difference(20.0)

# Spawn vehicles (batch spawn code omitted for brevity)
# Assume 'vehicles' is a list of spawned vehicle actors

for vehicle in vehicles:
    vehicle.set_autopilot(True, tm_port)

    # Make 10% of vehicles aggressive drivers
    if random.random() < 0.1:
        # Per-vehicle overrides: faster and closer following
        traffic_manager.distance_to_leading_vehicle(vehicle, 1.5)
        traffic_manager.vehicle_percentage_speed_difference(vehicle, -15.0)
        # Allow lane changes and ignore distance to other vehicles
        traffic_manager.auto_lane_change(vehicle, True)
        traffic_manager.ignore_lights_percentage(vehicle, 5.0)
    else:
        # Standard drivers keep global defaults
        traffic_manager.auto_lane_change(vehicle, True)

Related Pages

Implements Principle

• Principle:CARLA_simulator_Carla_Traffic_Behavior_Configuration

Uses Heuristic

• Heuristic:CARLA_simulator_Carla_Traffic_Manager_Sync_Mode