Ai for Game: whitehall mystery

Started by
6 comments, last by sefiroths 6 years, 1 month ago

Hi, I'm looking for suggestions for this game similar to ravensburger scotland yard

any deas is appreciated

 

https://images-cdn.fantasyflightgames.com/filer_public/78/b4/78b4b240-ec1d-416d-8486-970fb5a941c9/whitehall_mystery_rulebook_small_copy.pdf

thanks

Advertisement

Suggestions for what? 

Dave Mark - President and Lead Designer of Intrinsic Algorithm LLC
Professional consultant on game AI, mathematical modeling, simulation modeling
Co-founder and 10 year advisor of the GDC AI Summit
Author of the book, Behavioral Mathematics for Game AI
Blogs I write:
IA News - What's happening at IA | IA on AI - AI news and notes | Post-Play'em - Observations on AI of games I play

"Reducing the world to mathematical equations!"

What kind of algorithm should be batter for policeman?

it is like pacman where pacman is not visibile....

montecarlo tree search, neural network...

I haven't played the game and I didn't bother reading the rule book, but if the problem is chasing some hidden target through a graph, you should probably look into hidden Markov models. The short description is that your current understanding of where the target might be is a probability distribution over the nodes of the graph, and then two types of things might happen:

  1. Time elapses: A transition matrix describes the probability of moving from one node to another, and you should update your probability distribution with a matrix-vector multiplication.
  2. An "observation" occurs: If know the conditional probability of the observation given the target is at node N, for all possible values of N. You should update your probability distribution using Bayes' formula.

I don't know what kinds of decisions are made in that game, but it's likely that this way of thinking of the situation is helpful.

Also, look into occupancy maps as popularized by Damian Isla in Third Eye Crime.

 

https://www.youtube.com/watch?v=5lBye82qUog

https://www.gdcvault.com/play/1018057/From-the-Behavior-Up-When  (Start at 30:00)

Dave Mark - President and Lead Designer of Intrinsic Algorithm LLC
Professional consultant on game AI, mathematical modeling, simulation modeling
Co-founder and 10 year advisor of the GDC AI Summit
Author of the book, Behavioral Mathematics for Game AI
Blogs I write:
IA News - What's happening at IA | IA on AI - AI news and notes | Post-Play'em - Observations on AI of games I play

"Reducing the world to mathematical equations!"

"Occupancy maps" seems to be a rediscovery of hidden Markov models by the game industry. :) The math involved is exactly the same.

Hi, Thanks for the answers. The Game is turn-based and the "visibility " of the policeman is 1 adiacent  node of the graph and only if they search for clues... the killer should reach his destination in 15 moves. seem pretty hard to use this method, or am I wrong?

thanks

This topic is closed to new replies.

Advertisement