Title :
Neural Reinforcement Learning Controllers for a Real Robot Application
Author :
Hafner, Roland ; Riedmiller, Martin
Author_Institution :
Neuroinformatics, Osnabrueck Univ., Osnabruck
Abstract :
Accurate and fast control of wheel speeds in the presence of noise and nonlinearities is one of the crucial requirements for building fast mobile robots, as they are required in the MiddleSize League of RoboCup. We will describe, how highly effective speed controllers can be learned from scratch on the real robot directly. The use of our recently developed neural fitted Q iteration scheme allows reinforcement learning of neural controllers with only a limited amount of training data seen. In the described application, less than 5 minutes of interaction with the real robot were sufficient, to learn fast and accurate control to arbitrary target speeds.
Keywords :
control nonlinearities; iterative methods; learning (artificial intelligence); learning systems; mobile robots; multi-robot systems; neurocontrollers; velocity control; RoboCup; control nonlinearities; mobile robots; neural fitted Q iteration; neural reinforcement learning controller; speed controller; Automatic control; Control nonlinearities; DC motors; Learning; Mobile robots; Optimal control; Robotics and automation; Shape; Training data; Wheels;
Conference_Titel :
Robotics and Automation, 2007 IEEE International Conference on
Conference_Location :
Roma
Print_ISBN :
1-4244-0601-3
Electronic_ISBN :
1050-4729
DOI :
10.1109/ROBOT.2007.363631