Title :
A comparison of direct and model-based reinforcement learning
Author :
Atkeson, Christopher G. ; Santamaría, Juan Carlos
Author_Institution :
Coll. of Comput., Georgia Inst. of Technol., Atlanta, GA, USA
Abstract :
This paper compares direct reinforcement learning (no explicit model) and model-based reinforcement learning on a simple task: pendulum swing up. We find that in this task model-based approaches support reinforcement learning from smaller amounts of training data and efficient handling of changing goals
Keywords :
learning (artificial intelligence); model reference adaptive control systems; nonlinear control systems; robots; acrobot; direct reinforcement learning; model-based reinforcement learning; pendulum swing-up; Computational modeling; Control system synthesis; Control systems; Educational institutions; Force control; Jacobian matrices; Learning; Robots; State-space methods; Training data;
Conference_Titel :
Robotics and Automation, 1997. Proceedings., 1997 IEEE International Conference on
Conference_Location :
Albuquerque, NM
Print_ISBN :
0-7803-3612-7
DOI :
10.1109/ROBOT.1997.606886
