Posts from AI for Heroes Auto-translated

Hello everyone!

I work with optimal control and reinforcement learning. Given the breakthroughs in RL for chess and Go, I was wondering if it's possible to improve the AI in my favorite turn-based game – Heroes? I'd like to try out a couple of algorithms. Could someone point me in the right direction, which branches to clone and which classes to look at first in the code?

Thank you!
w

If you're interested, there's an open-source project called vcmi: https://github.com/vcmi/vcmi
And here's an attempt to rewrite the vcmi AI: https://github.com/vcmi/vcmi/tree/nullkiller-ai

The AI is still far from being playable in terms of "intelligence." There are bugs, and it also can't do many basic things.
Here's the conference: https://app.slack.com/client/T2JA4F4CX/C2J9QFUDS

The project is in a semi-frozen state, so don't expect quick responses. But the source code might be useful for your own purposes. If you're interested, feel free to join.

Thanks! I'll gradually delve into it.