I am researching different AI approaches and was curious what approach would be useful in my scenario.

Assume you are tiling a room. The tiles, and the room itself, can be any shape. In this room you could encounter N number of obstacles, such as a wall, or built-in. The goal is to layout the tiles, taking into account cutting into the obstacles mentioned above, along with the shape and dimensions of the destination room. This would have to account the shape, and measurements of said tile being placed onto the room.

Which AI approach would prove useful in this scenario?

  • $\begingroup$ Hi, that is an interesting question. It seems to be an optimisation question, but it is not clear what you are trying to optimise. For example do you want to minumise the number of tiles used (and if so, do you allow re-use of cut tiles if pieces can be re-used), or minimise the number of partial tiles, or have some more complex cost function? Also, if the tiles can be "any shape", do they need to be a tesselating shape, or even a single shape? $\endgroup$ – Neil Slater Feb 23 at 21:24
  • $\begingroup$ Hey Neil, yes I think reuse would nice however could be allocated as a waste ratio too, like always assume a throwaway for shapes manipulated to fit into the larger shape, which would in turn increase the tally of supply accordingly. $\endgroup$ – user3010406 Feb 23 at 22:16
  • $\begingroup$ To be clear, your loss metric is the waste ratio (total area of whole tiles used, divided by area of room), with an ideal score of 1.0? Could you perhaps state that in the question, and ideally mathematically (give an actual formula), although probably the formula is not an important part of any answer, just it is important that you are able to express a cost function in order to clearly make this an optimisation problem. Use edit to adjust the question $\endgroup$ – Neil Slater Feb 24 at 8:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.