Model-based and model-free reinforcement learning for visual servoing
- A.M. Farahmand
- Azad Shademan
- Martin Jagersand
- Csaba Szepesvari, Department of Computing Science; PI of AICML
To address the difï¬culty of designing a controller for complex visual-servoing tasks, two learning-based uncalibrated approaches are introduced. The ï¬rst method starts by building an estimated model for the visual-motor forward kinematic of the vision-robot system by a locally linear regression method. Afterwards, it uses a reinforcement learning method named Regularized Fitted Q Iteration to ï¬nd a controller (i.e. policy) for the system (model-based RL). The second method directly uses samples coming from the robot without building any intermediate model (model-free RL). The simulation results show that both methods perform comparably well despite not having any a priori knowledge about the robot.
Citation
A. Farahmand, A. Shademan, M. Jagersand, C. Szepesvari. "Model-based and model-free reinforcement learning for visual servoing". IEEE International Conference on Robotics and Automation (ICRA), pp 2917-2924, May 2009.Keywords: | Reinforcement Learning, Robotics, Vision, Visual Servoing |
Category: | In Conference |
BibTeX
@incollection{Farahmand+al:ICRA09, author = {A.M. Farahmand and Azad Shademan and Martin Jagersand and Csaba Szepesvari}, title = {Model-based and model-free reinforcement learning for visual servoing}, Pages = {2917-2924}, booktitle = {IEEE International Conference on Robotics and Automation (ICRA)}, year = 2009, }Last Updated: January 05, 2012
Submitted by Azad Shademan