Navigation vs ML Agents for airplane behavior

Hello, I’m developing a game with a friend and we’re trying to replicate some of the behavior of an airplane in an airport, basically:

  • Automated landing.
  • Automated movement when landed from A to B.
  • All of the above with other vehicles and obstacles in the way.

We’ve been analizing options and so far the 2 best fits we think are:

  • Navigation.
  • ML Agents.

In your experience, what would fit this use case the most, are there advantages/disadvantages to either? From my personal testing ML agents seems like more work up front but less work in the long run, while navigation could be done with a couple of nav mesh plains and stairs(?) for the plane to go up and down.

Thank you so much in advance, this is our first game so we’re still getting the hang of this.

my thought
animated landing is a position curve and speed curve, if kinematic.

So you're saying calculate all the points in a bezier curve, and just move from point to point?

or unity's AnimationCurve (you'll have to adjust values for landing start height)

that would much easier than navigation, but the airplane should look straight towards the runway...
you didn't mention your controller, and if you have ai flying implemented, then it's also not much more complicated to navigate towards the runway and land..

I used to have the player manually control the planes, but we changed the scope, now we want to tell the plane where to land, after it has spawned in a random position, looking at a random direction.