Can OptaPlanner be used to do physical layout problems?

Question

I'd like to use OptaPlanner (or something comparable that allows for constraint based optimization) to do vegetable garden bed layout. Ideally this would take into account the space needed per plant, which plants do well next to each other and which don't, and then also how long it takes for the plant to produce food, how much food, etc. The idea would be to help plan a full season where as soon as a plant has been harvested something else can be dropped into its place to max out the productivity and the continuity of food availability. Last, I'm hoping to create aesthetic rules which bias designs towards more pleasing layouts. For example symmetry & repeated (rhythmic) elements vs randomness and ease of harvesting large contiguous groupings.

I think most of these I can figure out how to create constraints in OptaPlanner, but I'm not sure how to represent the physical space and the proximity of each location to its neighbors. Would I break each bed into a set of grid cells, and then assign plants of different cell sizes to regions? Or maybe this needs to be turned into a graph representation? Are there any other physical layout example scenarios I can build off of?

Answer 1

At the moment, we do not have any examples for using OptaPlanner on a bin packing problem. (Which I think your situation could be mapped onto.)

Assuming you can separate your garden bed into a predefined set of fixed slots, in which a plant either does or does not fit, I think OptaPlanner could be used. These slots would then have a fixed set of geographic data, like which is close to which.

If, on the other hand, you can not subdivide the garden beds beforehand and would like the solver to do that for you as well, then the problem becomes much harder to solve. You are solving two problems, one of which is spatial - and you may actually find some success treating the two problems individually. First, find the ideal layout based on the number and type of plants you want to plant, and in the second step figure out which plants to put where.