Game Board
Game tree animation (including 4 stages)
How AI thinks
AI games usually have too many algorithms behind it and its “thinking” process is complex to get a concrete idea for players. Although there are many good adversarial search algorithms right now, it is still too abstract to connect those complex algorithms with the games we are playing together, also it is necessary to clarify how the game tree expand while keeping the environment as invariant. In this work, I will design an AI game tree visualization system which enable flexible visual adversarial search algorithm exploration by allowing the user to modify parameters of AI algorithm -- Monte Carlo Tree Search, playing against the AI and see the real-time animation. During the playing procedures, for each predicted step of AI, the system will visualize the game tree and the animation of its expanding and the corresponding properties of the AI algorithm. The system will deploy a simple game engine for a simple game, connect-4, and synchronize another interface for the game tree algorithm interactive animation. With the system, the user can explore to find better properties (for example iteration) of AI algorithm, also could find a better combination of parameters. On the other hand, for beginners, the system could work perfectly as a visualization tool to comprehensive the complex AI algorithm quickly. In this paper, I will introduce how to implement the system in details, including three parts: game engine implementation, game tree algorithms, interactive and real-time interactive animation and visualization.
Author: Sherry/Xueran Ma
Course CSE261 2020 Fall
Game Board
Game tree animation (including 4 stages)
Click Game executation file to start the app but please wait around 1 min to launch.
Specific instructions and its coresponding buttons are as the left picture above
In source folder, game is executable file (pls wait around 1 min to lauch the canvas), game.py is the source file
Source Folder