This application teaches a learning agent built with the GEGELATI library how to keep a pendulum at an equilibrium position.
The build process of applications relies on cmake to configure a project for a wide variety of development environments and operating systems. Install cmake on your system before building the application.
- Copy the
gegelatilib-<version>
folder containing the binaries of the GEGELATI library into thelib
folder. - Open a command line interface in the
bin
folder. - Enter the following command to create the project for your favorite IDE
cmake ..
. - Open the project created in the
bin
folder, or launch the build with the following command:cmake --build .
.
The folder src/CodeGen contains an example of use case for the code generation feature of GEGELATI. There are 4 targets for this example:
-
pendulumCodeGenCompile
: Import the TPG_graph.dot and launch the code gen to generate the sources files. If you want to run this target you need to set your working directory as the current build directory of your build system. You can use the following variable$CMakeCurrentBuildDir$ . -
pendulumCodeGenGenerate
: A custom command to execute the previous target (after it is compiled) -
pendulumCodeGenInference
: Uses the generated file and link them with the learning environment of the directory èsrc/Learn
. This target depend on the previous, so building it will automatically trigger a build of the two previous. -
pendulumTPGInference
: Import theTPG_graph.dot
and run it within the pendulum learning environment, in the exact same condition as within thependulumCodeGenGenerate
target. This target enables comparing the identical behavior of the generated code and the original TPG.