GridworldQLearning.hpp

Go to the documentation of this file.

#ifndef SRC_APPLICATION_GRIDWORLDQLEARNING_HPP_
#define SRC_APPLICATION_GRIDWORLDQLEARNING_HPP_

#include <vector>
#include <string>

#include <opengl/application/OpenGLEpisodicApplication.hpp>
#include <opengl/visualization/WindowCollectorChart.hpp>
using namespace mic::opengl::visualization;

#include <types/Gridworld.hpp>

namespace mic {
namespace application {



class GridworldQLearning: public mic::opengl::application::OpenGLEpisodicApplication {
public:
    GridworldQLearning(std::string node_name_ = "application");

    virtual ~GridworldQLearning();

protected:

    virtual void initialize(int argc, char* argv[]);

    virtual void initializePropertyDependentVariables();

    virtual bool performSingleStep();

    virtual void startNewEpisode();

    virtual void finishCurrentEpisode();


private:

    WindowCollectorChart<float>* w_chart;

    mic::utils::DataCollectorPtr<std::string, float> collector_ptr;

    mic::environments::Gridworld grid_env;

    mic::types::TensorXf qstate_table;

    mic::configuration::Property<float> step_reward;

    mic::configuration::Property<float> discount_rate;

    mic::configuration::Property<float> learning_rate;

    mic::configuration::Property<float> move_noise;

    mic::configuration::Property<double> epsilon;

    mic::configuration::Property<std::string> statistics_filename;

    std::string streamQStateTable();


    float computeBestValue(mic::types::Position2D pos_);


    mic::types::NESWAction selectBestAction(mic::types::Position2D pos_);


    long long sum_of_iterations;

    long long sum_of_rewards;

};

} /* namespace application */
} /* namespace mic */

#endif /* SRC_APPLICATION_GRIDWORLDQLEARNING_HPP_ */