Decision Interval benefits

[testContributor12]

What are the actual benefits of using a Decision Interval that is more than 1?

As far as i understand if we assign the Decision Interval to 5 , then it will make 5 different actions, and after they are complete the agent will get the new observations and make another 5 actions.
eg. action1:movetox , action2:rotate and moveY, etc..

I feel like i am wrong somewhere ,because in that case using a Decision Interval of 1 will always lead to a more accurate calculations.

[mbaske]

With decision interval set to 5, the agent actually repeats the same action 5 times before making the next decision step. You could implement this behaviour using on-demand-decisions and request a new decision every 5th time the agent's FixedUpdate method executes. Setting the decision interval to greater than 1 is just a convenient way for spreading out decisions without having to write your own interval code. Longer decison intervals can be beneficial if an agent doesn't need updates on every frame, e.g. for deciding which direction to move in.

[testContributor12]

I see, thank you very much for the explanation!

[github-actions]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.