Hello, welcome to our CompSci 175 AI Project - iKun AI’s homepage!

Here is our cute agent’s prototype Aiku! drawing

IKun AI is an Artificial Intelligence Project on Deep Q-Learning with Neural Network. The project is designed to let an agent play in a set of pre-built Minecraft 2D mazes. The agent is able to see the voxels in front of its sight, which is similar to what a normal player can see in Minecraft. The agent’s goal is to get out of the maze as fast as possible using the limited resources that are given to the agent.

Our agent Aiku is a cute pullet good at singing, dancing, rapping and playing basketball. One day, she opened a box and she jumped into the Minecraft World. She was stuck in a very dangerous maze. Although with wieht the help of the force, a map appeaered in her thoughts, he doesn’t know where she is in the map.

That is why we give her instructions by our intelligent algorithm with Deep Q-learning so that she can explore the maze firstly and guess exact location of her sight.

drawing

drawing

drawing

drawing

drawing

drawing

drawing

drawing

In this Minecraft World, Aiku finds herslef at somewhere of a maze. She receives a map of the maze from a mysterious person. However, she does not know where exactly she is in the map, which means she needs to determine the position of herself in the map. Also, this maze is full of fatal flowing lava, so Aikun really wants herself to get out of the maze as fast as possible. Here are some parameters of the maze, it is a 21 x 21 square. Aiku starts at a block called “emerald_block,” and ends at a block called “redstone_block.” Aiku is walking on diamond blocks. The environment around her is extremely dangerous since it is full of lava. What Aiku can do in one turn is either detect how many blocks in front of her are walkable, turn left, turn right or move 1 block forward. Since she already have the map, what she needs to do is just locate herself. After that, Aiku is smart enough to go to the ending block of the maze in the fastest path.

Source code: https://github.com/yuexintonychen/iKun-AI

Reports: