Sciweavers

Share
ICRA
2002
IEEE

The Necessity of Average Rewards in Cooperative Multirobot Learning

9 years 4 months ago
The Necessity of Average Rewards in Cooperative Multirobot Learning
Learning can be an effective way for robot systems to deal with dynamic environments and changing task conditions. However, popular singlerobot learning algorithms based on discounted rewards, such as Q learning, do not achieve cooperation (i.e., purposeful division of labor) when applied to task-level multirobot systems. A tasklevel system is defined as one performing a mission that is decomposed into subtasks shared among robots. In this paper, we demonstrate the superiority of average-reward-based learning such as the Monte Carlo algorithm for task-level multirobot systems, and suggest an explanation for this superiority.
Poj Tangamchit, John M. Dolan, Pradeep K. Khosla
Added 15 Jul 2010
Updated 15 Jul 2010
Type Conference
Year 2002
Where ICRA
Authors Poj Tangamchit, John M. Dolan, Pradeep K. Khosla
Comments (0)
books