akshaynarla/CarlaSumoArtery-CoSimulation_clone

CarlaSumoArtery-CoSimulation

Requirements

Environment

• Desktop computer with a monitor (not tested on cloud or remote desktop)
• Linux: Ubuntu 18.04 or 20.04

Simulators to Download

• CARLA v.0.9.[11-13]
• Artery
• Sumo 1.8.0
• Omnetpp 5.6.2

Installation

Pre-requisites

• Dependecies: pip3 install psutil pyproj
• Set Omnetpp
  • ./configure WITH_TKENV=no WITH_QTENV=no
  • make cleanall
  • make MODE=release
• Ensure CARLA and Artery and Sumo are installed and working

Add the following folders

• In Artery, add the folder named "artery/src"
  • Note: 03/2022, the file Core.cc is no longer needed if you use the lastest version of artery
• In CARLA, add the folder named "carla/Co-Simulation"
• Add the folder named "simulation_modules" next to the carla folder

Testing

• Run server mode carla: /Path/To/Carla/CarlaUE4.sh
• Connect as a Carla client:
  • cd to parent folder of carla and simulation_modules
  • python3.7 synchro_client.py
    • Note: You may have face some permission denied error on an artery file named run_artery.sh. Ensure to allow executing files as a programm (Right click on the file > Properties > Permissions > Execute (check the box)
• command example :
  • python3.7 synchro_client.py --color-agents --color-cams --num-clients 2 --start-artery --artery-build-path /home/med/artery/build
  • Run "python3.7 synchro_client.py -h" for some details about possible configurations

Adding new security features

How it works

• The synchro_client.py script uses carlas's "run_synchronization.py" to synchronize carla and sumo and then in the simulation loop looks for a connection from Artery simulation to start recieving messages. The script also looks for defined attackers and detection mechanisms and runs them in the same loop and plots corresponding visualization in carla.

new detector

• Adding a new detector should be done as follows :
  • Create a class for the detector (subclass of Detector)
  • Implements the function check().
    • check() takes as argument the current step cam messages ( which defines wether the detector is a global one or a local one depending if it receives all messages or just specific messages received by one vehicle )
    • check() then retuns the ids of detected misbehaving vehicles.

new attack

• Adding a new attacker should be done as follows :
  • Create a class for the attacker (subclass of Attacker)
  • Reimplement the is_ready() function if necessary (if not, the attacker will perform an attack every simulation step).
  • implement the perform_attack() function.
    • perform_attack() function must create a carla object in the position where the attack will happen.

Misc

Reference

If you use our work, please cite us!

@inproceedings{bouchouia2022simulator,
  title={A simulator for cooperative and automated driving security},
  author={Bouchouia, Mohammed and Monteuuis, Jean-Philippe and Labiod, Houda and Jaballah, Wafa Ben and Petit, Jonathan},
  year={2022}
}

Authors

• Mohammed Lamine Bouchouia
• Jean-Philippe Monteuuis
• Houda Labiod
• Ons Jelassi
• Wafa Ben Jaballah
• Jonathan Petit

Contact:

A question? Send us an email: mohammed.bouchouia@telecom-paris.fr