| We're trying to work on planning. The hope is that we can avoid a great deal of search by reasoning about goals and actions. Here's a simple three-in-a-row eyespace problem: ..... ..... BBBBB wwwww w...w Our system is capable of reasoning that playing in the center is the only way for white to make two eyes. The problem with this is that the effects of some actions depend on the state of the board. Consider this situation: ..... ..... BBBBB wwwww w.B.. The system realizes that playing to the right of the black stone makes an eye in the corner, but it doesn't realize it can capture the black stone for a second eye. The problem is that the effect of capturing the black stone depends on playing BOTH moves. Thus, simple STRIPS operators (with their add and delete lists) are not sufficient. Any ideas? Peter Drake Assistant Professor of Computer Science Lewis & Clark College |
_______________________________________________ computer-go mailing list computer-go@xxxxxxxxxxxxxxxxx http://www.computer-go.org/mailman/listinfo/computer-go/