OpenCDA Overview
OpenCDA is the scenario orchestration layer of CAVISE. It connects automated
driving logic to the CARLA world and can optionally synchronize the same
experiment with SUMO and Artery. A scenario describes the world, vehicles,
roadside units, sensors, driving behavior, communication services, metrics,
and attacks in YAML. opencda.py loads that configuration and runs the
simulation loop.
For setup and the first run, follow Installation and First Simulation.
What OpenCDA Provides
Area |
Available functionality |
|---|---|
Scenario orchestration |
Deterministic CARLA worlds with configurable maps, weather, simulation steps, vehicles, RSUs, destinations, background traffic, and random seeds. |
Automated driving |
Localization, camera and LiDAR perception, map management, safety monitoring, route and trajectory planning, PID control, CARLA autopilot, and platooning. |
Co-simulation |
Bidirectional CARLA–SUMO synchronization for traffic and CAPI-based communication with Artery network simulations. |
Cooperative applications |
Vehicle and RSU behavior services, including state publication, movement requests and control, and AIM client/server workflows. |
Cooperative perception |
OpenCOOD-based multi-agent perception with visualization, prediction export, and configurable evaluation metrics. |
Security experiments |
Declarative attacks against behavior services and AdvCP attacks against cooperative perception pipelines. |
Evaluation and data collection |
CARLA recording, sensor data dumping, structured logs, and metrics for localization, driving behavior, platooning, and cooperative perception. |
Scenario Demonstration
Cooperative Perception Demonstration
The v2xp_datadump_town06_carla scenario demonstrates cooperative
perception with a connected vehicle and roadside infrastructure. The first
view shows the fused detections in the CARLA scene. The second shows the same
run from the bird’s-eye-view visualizer.
