This introduction will give you an overview how to install, parametrize, and execute the falsification tool for Adaptive Cruise Control (ACC) systems which is based on the approaches introduced in M. Koschi, C. Pek, S. Maierhofer, and M. Althoff: Computationally Efficient Safety Falsification of Adaptive Cruise Control Systems, in Proc. of the IEEE Int. Conf. on Intelligent Transportation Systems, 2019.
For the execution of the falsification tool you need at least Python 3.6 and the following modules:
- commonroad_io >= 2019.2
- matplotlib >= 2.5.0
- numpy >= 3.1.0
- imageio >= 1.16.4
- scipy >= 0.16.5
- ruamel.yaml >= 1.3.1
The usage of the Anaconda Python distribution is recommended.
You can install the required Python packages with the provided requirements.txt file (pip install -r requirements.txt).
Additionally, you need the CommonRoad vehicle models which must be added to your Python interpreter path.
The three main files are falsification.py, config.yaml, and review.py:
1. config.yaml: This file allows to adapt all parameters used by the falsification system, e.g. ACC vehicle parameters, simulation parameters, leading vehicle parameters).
2. falsification.py: This file starts the falsification, e.g., by executing it within the command line ("python falsification.py > solution.out"). If the storage of solution files is activated in config.yaml or if a valid solution is found, solution files are generated with the initial setup (config_xyz.yaml) and the lead vehicle trajectory (trajectory_xyz.pkl), where xyz in the name of the generated files corresponds to a concatenation of the current date and time. The initial state of the ACC vehicle corresponds to the values defined in config.yaml (s_init, v_init). The initial state of the leading vehicle is randomly set in front of the ACC vehicle according to the parameters in config.yaml (s_safe_init_backward, delta_s_init_min, delta_s_init_max).
3. review.py: This file allows to execute a stored initial setup and leading vehicle trajectory. The script plots different profiles of the leading and ACC vehicle and stores a CommonRoad scenario. You must provide an input parameter to the script (e.g., "python review.py xyz", where the solution files are called config_xzy.yaml and trajectory_xyz.pkl).
The folder acc contains different ACC systems. If you want to add your own ACC system, you must implement the ACC system corresponding to the interface in acc_interface.py. Additionally, you must add your ACC system to the dictionaries in config.yaml and to the create_acc_vehicle_param() function and the ACCController enum, both located in configuration.py.
A detailed documentation can be found under doc/build/html/index.html.