Robot Control Using Model-Based Reinforcement Learning With Inverse Kinematics