GeorgeNagel / simpleplanning

A planning library in Python
7 stars 1 forks source link

Question about Probabilistic actions #1

Open ivan133 opened 9 years ago

ivan133 commented 9 years ago

Hello! Is't not a real issue, just a question. Today I'm pretty interested in writing text game, with probabllistic actions. In the readme it is said that it is in ToDo. Do you have any success in this direction? Do you plan to develop this repository further? Maybe you have some architectural ideas you can share?

GeorgeNagel commented 9 years ago

I don't have plans to develop this repository any further for the near future, at least not for the next six months or so, but here are my current thoughts regarding probabilistic actions:

If the probability for the success of each possible action is known (given a known state of the game world) then you should be able to calculate the probability of successfully reaching some end game world state just by multiplying the individual success probabilities of each branch on the planning tree.

For more realistic effect, each character may have a different understanding of the current state of the game world that affects their probability calculations (one character believes they'll be able to climb a wall because they think that there's a ladder, but when they get there they discover that the ladder is no longer there.) Additionally, each character may estimate the success probabilities to have different values; an overconfident character may always estimate that the probability of success for their actions is higher.

If you're dealing with intelligent characters, then it may be interesting to not always pick the path with the highest probability of end success. For example, a more emotional or short-sighted character might pick an action towards their goal that has the highest probability of success at that first step. A balance between characters who choose near-term success vs. long-term success may create a more interesting game for the player.

Those are my thoughts for now. Best of luck on your text game!