|
Title:
|
ACTION SELECTION FOR MULTI-SMDP BASED ROBOTS WITH DYNAMICALLY PRIORITIZED GOALS |
|
Author(s):
|
Niko Prüßner, Willi Richert, Christoph Rasche, Achim Rettberg |
|
ISBN:
|
978-972-8939-23-6 |
|
Editors:
|
António Palma dos Reis and Ajith P. Abraham |
|
Year:
|
2010 |
|
Edition:
|
Single |
|
Keywords:
|
Strategy prioritization, learning, robot, multi-SMDP |
|
Type:
|
Full Paper |
|
First Page:
|
75 |
|
Last Page:
|
82 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
This article considers the problem of how a robot can improve its overall performance by a smart scheduling of its diverse goals in the presence of dynamic goal preferences. These goals are efficiently approached by executing the preferred actions of corresponding SMDPs. Several methods to conduct the goal selection are presented. The effectiveness of the developed approaches is evaluated by simulation on different scenarios. One of the methods, "CR", shows an optimal behavior and can thus be interpreted as an upper bound. This method is further improved by the method CR-PS", which succeeds to reach this upper bound in a fraction of the original runtime. |
|
|
|
|
|
|
