|
|||
|
My Master Thesis "The Reinforcement Learning Toolbox, Reinforcement Learning for optimal control tasks" is finished since June 2005, it contains a comprehensive description of the class system of the RL toolbox. The use of RL for optimal control tasks is explained and many different algorithms are introduced. These algorithms are among others continuous time RL, continuous actor critic learning, Residual and Residual Gradient algorithms, Policy Search algorithms like CONJMDP and PEGASUS. These algorithms are tested and compared on three different benchmark tasks which are the Pendulum, Cart-Pole and Acrobot swing up task. You can download the thesis here. |
