SuperTuxKartAI - Deep Learning - Summer 2020

Term project for Deep Learning at the University of Texas at Austin. Task is to train a neural network to play SuperTuxKart ice-hockey (basically Rocketleague).

Getting Started

Tensorboard

In a seperate terminal, Run : tensorboard --logdir tmpLogs --reload_interval 1

To train this ish:

1. Unzip data.zip into the project directory
2. Increment the --run env variable found on line 116 of agent/train.py
3. Run: python -m agent.train <add env variables here OR more on that below>

Note Refer to agent/train to see the hyperparameters that you can tweak...also feel free to "hard-code" those parameters in the bottom of the file as well.

To play da game with 2 Players:

1. Run python -m tournament.play agent agent -s gameplay/

Note After this finishes running, there should be a .mp4 file in gameplay for each of the 2 players.

If you want to generate new data.

TLDR:

1. Run

python -m tournament.gen_train_data random_agent AI random_agent AI -s data\train -f 1600

Note There's a lot of different ways to do this but couple of tips to follow.

• make sure that whatever value in -f multiplied by the number of "random_agent"s is a multiple of 64, 128, 256 (our typical batch sizes)

Hit Lewai up too cuz there's a couple of commands and it could get messy/complicated.

=================================================================================================================

Project Description

starter code

The final project is open-ended. The task is to program a SuperTuxKart ice-hockey player.

In agent/player.py, implement the class HockeyPlayer. This class provides an act function, which takes an image (in numpy format) and the state of the player (ground truth position and camera parameters). You should return an action (steering, acceleration, ...). random_agent contains an example of a random hockey player than just drives straight with minor random perturbations.

We also provide you with a hockey tournament script. Run it with

python -m tournament.play player1 player2 ... where playeri is any agent you code up (e.g. agent) or AI to use the game AI to play. The teams of the agents alternate. For example

python -m tournament.play random_agent AI random_agent AI sets up a match between two random agents and two AI agents.

Use the --save flag to store your agent's output.

The rules

This project is completely open-ended, however, there are a few ground rules:

1. You may work in a team of up to 4 students.
2. Teams may share data, but not code.
3. You may organize test races between each other. However, make sure to delete your agents from the shared filesystem after the tournament (no sharing of code).
4. In our grader, your agent will run in complete isolation from the rest of the system. Any attempt to circumvent this will result in your team failing.
5. You may use pystk and all its state during training, but not during gameplay in the tournament.
6. Your code must be efficient (< 100ms / step) on a CPU. There is no GPU access during testing.

The tournament

We will play an ice hockey tournament between all submissions. Two of your agents will play an opponent. Your agents will run in the same process and may communicate with each other. Each game will play up to 3 goals, or a maximum of 1200 steps (2 minutes). The team with more goals wins. Should one of the teams violate the above rules, it will lose the game. Should both teams violate the rules, the game is tied.

Depending on the number of submissions, we will let all submissions play against each other, or we will first have a group stage and then have the top 10 submissions play against each other. The submission with the most victories wins the tournament. The goal difference breaks ties. Additional, matches further break ties.

Grading

40pts originality of the idea. How many things did you try? Do you understand how your strategy works? 30pts average number of goals scored per game. Linearly from 0 to 1. We will compute the average against the game AI or the rest of the class and use the better of the two. 30pts writeup. Structure of the writeup, exposition, how well is the idea motivated and explained. 15 pts extra credit for the top 3 teams in the competition (15pts winner, 10pts second, 5pts third).

Hints

Test your code on a Linux machine (either CS lab or colab). We will organize several mock tournaments before the final tournament. Make use of them.

Submission

Once you finished the project, create a submission bundle using

python bundle.py agent [YOUR UT ID]

Make sure to place your writeup in the writeup directory, and remove any non-essential files from it.