A game playing robot that can learn from experience

Abstract

We present a new approach for online learning an x-o game strategy by humanoid robot ldquohoap-3rdquo*. No preset data for game playing are provided in advance. The proposed system mechanism simulates human decision making to carry out the online game learning. ldquohaop-3rdquo autonomously gains experience needed for learning the game strategy from the human partner. The more intelligent human partner, the faster humanoid robot ldquohoap-3rdquo learning.