2048 AI BOT

An AI bot for the game 2048.

Made for the competition of NTU Taiwan 2048 BOT Competition in my second half semester of first grade.

Build

(current support windows and linux)

AI version

$ make

$ ./bot2048

Manual play version (Use W,A,S,D to control)

$ make manual

$ ./manual2048

Build with docker

• Compile in Ubuntu docker manually

  1. Mount physical directory

     $ docker run -it --name test2048 -v "$(pwd)":/app ubuntu:latest

  2. Install compilation tool

     $ apt-get update

     $ apt-get install -y g++ build-essential

  3. Compile (using instructions above)

     $ cd /app

     $ make

  4. Rerun the container (after you stop the container)

     $ docker start -i test2048

  5. (Optional) Save the container using export

     $ docker export test2048 > test2048.tar

  Pros: The package is small and clean (only about 158 MB for g++ compiler and 62.7MB for build-essential)

  Cons: Too many manual steps (we need to install everything and run the container and then run the program by ourself)

  To Test:

  1. Commit the changes of container to build an new image
  2. Make a Dockerfile (i.e. Build an image only include our essential c compile tools) based on our manual installation (Act like my second method but take less space)

• Build a image based on gcc and compile it automatically in it and save the image

  1. Dockerfile:

     • AI version

        FROM gcc:4.9
        COPY . /2048-AI
        WORKDIR /2048-AI
        RUN make
        CMD ["./bot2048"]
       

     • Manual play version

        FROM gcc:4.9
        COPY . /2048-AI
        WORKDIR /2048-AI
        RUN make manual
        CMD ["./manual2048"]
       

  2. Build image

     $ docker build -t ai2048:v1 .

  3. Run image

     $ docker run -it --rm --name test ai2048:v1

  Pros: Very quick (after build an image, each time we want to execute program just need to run the container)

  Cons: The new image we built is as big as gcc image (1.37GB) (so it take double space in computer because of two images exist)

• Cross compile using gcc image and run it in a clean ubuntu image

  1. Compile using gcc image (it will save in the current directory)

     $ docker run --rm -v "$PWD":/usr/src/ai2048 -w /usr/src/ai2048 gcc:4.9 make

  2. Run in ubuntu container

     $ docker run -it --rm --name 2048AI -v "$PWD:/app" -w /app ubuntu:latest ./bot2048

  Pros: Very small and easy to maintain

  Cons: Haven't thought one

Author

• David Lee
• Tom Chang

Notes

Doxygen

• Where should doxygen comment blocks place

Docker

• Using C++ with docker engine
• Difference between docker save and export
• Export and import docker

VS Code

• C/C++ for VS Code