CONVINCE model2code Documentation
The model2code tool is developed as part of the CONVINCE project and is used to generate skill-level code from the high-level SCXML model of a skill.
Model2code can be used when the SCXML model representing the state machine of the skill is executed using an SCXML compiler. To handle interfacing between the skill and the rest of the system, there is the need to write the corresponding code. Model2code allows to generate the code for the skill, based on a given template. Allowing the user to focus on the high-level design of the system.
Program inputs:
An SCXML model that describes the behavior of the skill,
An XML file that outlines the full model of the program, including the behavior tree, skills, and components used in the system.
An XML file that specifies the interfaces between the behavior tree and skills, as well as between skills and components.
A directory containing templates for the files to be generated.
Program outputs:
A C++ file containing skill code.
A header file declaring the skill.
A main C++ file that includes the main function of the skill.
An SCXML standard file containing the compilable SCXML code of the skill.
A package.xml file with information about the ROS package.
A CMakeLists.txt file that incorporates the generated files.
Required parameters:
--input_filename: The path to the SCXML file that describes the behavior of the skill.--model_filename: The path to the XML file that describes the full model of the program.--interface_filename: The path to the XML file that describes the interfaces used.
By default, the program generates the code in the same directory as the SCXML file specified by the --input_filename parameter. However, you can select a different output directory by using the --output_path parameter.
Additionally, the program uses files from the templates directory by default to generate the code, but you can specify a different directory with the --templates_path parameter.
The --log_mode parameter can be used to enable verbose logging.
The generated skills are based on a behavior tree structure. Skills corresponding to condition nodes will have a ROS2 tick service, while skills corresponding to action nodes will have both tick and halt services.