Complete Information Games

In game with complete information, each player perfectly knows the situation of the game. There is a winning strategy in (2-person zero-sum finite static) game with complete information, but we can acquire the strategy for only small size games. We discover a winning strategy of cubic 4-line tic-tac toe by using inductive logic programming.

Incomplete Information Games

In game with incomplete information, a player cannot perfectly know the situation of the game. For some games using playing cards, we research player agent based on reinforcement learning, and an algorithm for estimating player’s hands from his/her actions.

Discrete 3D puzzles

In order to realize a virtual 3D puzzle, effective 3D interaction techniques are required. Once such techniques are implemented, not only virtual puzzles similar to real ones, but also physically impossible puzzles can be designed.

Continuous 3D puzzles

The search space of a puzzle ring is a 6 dimensional continuous space. The solving such a puzzle requires an effective path planning algorithm. We have applied Rapidly-exploring Random Trees (RRT) to puzzle rings.