I have a steady hex-map and turn-based war game featuring WWII carrier battles. I would like to improve the fixed policy for the AI using reinforcement learning. I have some beginner's questions, which I will try to spread across several posts.

The game use fog of war.

  • Information about the game is zero at the beginning
  • and is slowly disclosed (approximate position and composition of naval task forces)

What is the best approach to deal with that?

Partial info on blue naval forces. Others are not visible on map

Partial info on blue naval forces. Others are not visible on map

  • $\begingroup$ This is a partial observation problem. Search for "partial observation" in an academic search and in this exchange. ... Quick, sidebar: The movie of the same name, Fog of War, is really interesting from a historical perspective. Robert McNamara's end of life confession sounds much like that of a sociopath. $\endgroup$ – han_nah_han_ Dec 24 '18 at 6:31

The problem which you want to solve is Reinforcement Learning with Partially Observable Markov Decision Process. I recommend you taking a look at papers which work in this formalism. For example, papers related to dota bot created by OpenAI. You can start here: https://openai.com/five/

| improve this answer | |
  • 1
    $\begingroup$ There are other proposed solutions to POMDP you may want to research and list here too. On this AI site, the more comprehensive answers seem to get more attention since this is appropriately more design and approach oriented than SO and some of the other SE sites. People often post here to check for options. $\endgroup$ – han_nah_han_ Dec 28 '18 at 3:04
  • $\begingroup$ @han_nah_han_ Unfortunately, I don't know everything in RL and just writing about what I know to give the poster some pointers. I hope somebody will come and write much more exhaustive answer. $\endgroup$ – Konstantin Solomatov Dec 28 '18 at 3:48
  • $\begingroup$ If everyone knew everything in RL, then it would work better for more usage cases, but it doesn't, so don't feel alone in that. $\endgroup$ – han_nah_han_ Dec 28 '18 at 3:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.