Hope that I can describe my question well with my limited math knowledge.
Imagine a gaming scenario. I have a tank that moves on a fixed-size map. And its coordinate will be sampled constantly every T milliseconds. Once I decide to attack a target. A mouse clicks on a target's coordinate with (Xt, Yt) will be recorded. At this point. I should have data of
- Target coordinate: (Xt, Yt), and
- A set of a tank's moving coordinate [(x1,y1),(x2, y2), (x3,y3).....(xn,yn)] (Length is fixed with N chronologically)
Is there any mathematical model that could take this training dataset set of object's movement coordinates as input and the target's coordinate (Xt, Yt) as output, and be trained to predict the target coordinate with only the tank's movement if a fire event is detected (predict (Xt, Yt))?
I guess the use case is also applicable to a 3D environment.