artificial intelligence - Random Decision Algorithm -

I am creating a simple AI and I am really new in this area. I need an algorithm to make decisions based on certain criteria; But with a little bit of randomness. What I'm doing so far is generating a random number and based on the different values ​​I found; Take different performance paths I somehow think that there is a better way to talk like this, can you give me some hints?

I do not think there are "better way" than you, because the problem The description of is quite unclear, and whether it is actually an artificial intelligence problem or the simulation problem is not clear.

For the part of the decision, a simple approximation system based on its underlying understanding of behavior based on the model based on) that is the sequence of conditional statements, possibly with some random factors, You can have a weighted average system to differentiate, which chooses many options, ranks them according to perceived quality, then being favoritism to better values. Chooses an option on Drichchik. (It is called roulette wheel selection or in genetic algorithm circles, but also very useful outside of GAs.)

For the simulation part, you usually want to model this process and then more subtle Want to present the randomness for the model? Parts of the process, it clearly means defining your input (backhand power, ball direction, ...) and your output (ball travel vector?) And How they are related to each other means that you give some output of known values, and then you can modify them with your random number generator. If you are trying to model a person's choice rather than just a physical simulation, then things get more complicated, because often takes priority over all other inputs.

A simple example I might start is that the player raises AI where he wants to hit the ball. I calculate the desired direction of the ball in the form of a vector which sends the ball exactly where the plan is made. The system then calculates a difficulty score of 0% to 100% on the basis of the direction of the ball, the speed, the ability of the player, etc. I generate an 'inaccuracy' launcher which generates a random number between 0% and difficulty%, which means that a perfectly easy shot will be completely accurate, while a half-hard shot 0% and 50% accurate Will be different between being. Then I calculate a random normal unit-length 3D vector, which adds to the impurity value, and in the desired direction. This means that more difficult shots will attempt to fly ahead of the goal.