Project Talos

Week 13 - Design Expo, EECS 494 Showcase, and Final Reflection

With this final devlog, I go over the Design Expo and the EECS 494 + EMU Showcase. I discuss the feedback I received over both events and their implications. Finally, I reflect on the project as a whole.

The Fall 2020 CoE Design Expo

The College of Engineering Design Expo was on Thursday, December 3rd. Over the four hour period of the event, a total of two people showed up to my presentation. This was very disheartening, to say the least, however the two people who did come showed a fair bit of interest in my research. I suspect the reason for the low attendence was due to the College of Engineering not announcing or advertising the event in any capacity. However, it is what it is.

The slides I presented at the event can be see here. They should be fairly straightforward to follow and contain many visuals. Hopefully, anyone who stumbles across this research in the future can find them helpful.

The Fall 2020 EECS 494 + EMU Showcase

Austin suggested we submit our research projects to the EECS 494 + EMU Showcase if we wanted to them to get more attention. As my research involved making an enjoyable game using machine learning, this was the perfect opportunity to see how “fun” my game really was. As such, I presented my project on Tuesday, December 8th at the showcase. I got a fair bit of people to check it out and play my game. Some were even nice enough to give me feedback.

Noteworthy Comments:

  • “I had a lot of fun reading through your slides and figuring out the game based on that! Really like the concept, and would not have been able to tell that the AI was ML based if I didn’t know it already… You did a great job, love the aesthetics.”

  • “This was a really interesting game and made me learn new ways to avoid the birds each play. It’s an ambitious project to make enemies that get better at the game as you do, but it turned out well. Sort of a temple run style game that’s really hard.”

  • “Really cool game and great tutorial. I can see this game can be made into a endless game like temple run. I really like the idea to have bird drinking your nectar while you run. The behavior of the birds are also great and realistic.”

Based on those and other comments, I’d say that the prototype game was a success. It was fun enough that people enjoyed it and the ML Agents were competent enough that people couldn’t even tell they were machine learning based.

Final Reflection

Looking back on this project, I can say although it had some rough periods, it was generally a success. I can look back at this project and say I am proud of what I was able to accomplish during the span of a semester. I was able to accomplish the goals I laid out for myself.

Accomplished Project Goals:

  1. I managed to create a fun and engaging game that used machine learning agents as a central game mechanic.
    • Many individuals found the game interesting and fun. Some even went as far as to say they wouldn’t have known the game used machine learning agents if it wasn’t mentioned.
    • I managed to show the possibility and feasibility of such a task.
  2. I documented the entire process along the way.
    • I covered everything from initial training to a final playable game.
    • I documented the problems I faced and the solutions I discovered along the way.
  3. I created a guide for making an ML Agent centered game from scratch.
    • I presented a straight forward approach to get from learning to use Unity ML Agents Toolkit to a final game using complex agents as a central game mechanic.
    • I covered tips for avoiding general issues.
    • I described an approach to make more complex agents.
  4. I provided paths to further research.
    • I briefly covered some avenues for future research discussing their benefits to academia and the difficulty and creative/intellectual freedoms involved.

In the process, I learned more about machine learning and created a game I’m happy to show to others. I can say I contributed to academia and the machine learning field. As such, I think it’s safe to close the door on Project Talos.

Thank you for everyone who helped along the way. This wouldn’t have been possible without you.