Can you elaborate? It seems to me that unless the screens are that far outside that they are where the target object is, two people that are offset laterally wrt the target object would have to be displayed something that's offset on the screen.

Imagine two passengers seating in rows r10 and r11, looking at the target T

A. You need to know where a passager's eyes are to display the POI in the right place. Even if each rows gets their own and only screen you'll need to account for their head vertical position (different people are different height) and movement, hence the eye tracking.

B. If you share a window between multiple people you end us with a POI mess with informations displayed multiplied by as much passengers in the bus.

   |- r9  -|w9
   |       |w9
   |- r10 -|w10   T
   |       |w10
   |- r11 -|w11
   |       |w11
   |- r12 -|w12

IMHO the only practical way is with personal headsets like [0] but then you don't need a bus: just use your foot or any transportation: it's AR and not VR.

0 https://www.youtube.com/watch?v=SpoLdQpPcAc

So is it an actual moving bus or just a simulation of one? I have not heard of the concept before