I am the developer of an online service for computer-aided design of irrigation systems. The working principle is somewhat similar to AutoCAD (or any CAD in general), but it is made specifically for irrigation purposes. I have a large database of custom projects, most of which have real-world applications. I want to make an assistant/copilot for my service, preferably neural network based, that with the push of a button will automatically give the user the possible/best sprinkler placement in their project.
I see this as a neural network that will get the coordinates of a particular lawn (which is a graph) via the API and then generate a possible sprinkler placement based on parameters such as sprinkler type/ lawn size/water flow rate and/or will use the user’s projects as a dataset to get weights or other performance estimates. As you have probably already realized, my knowledge in this area is very superficial, so I can’t wrap my mind around what instruments I should use for such a case. Can TensorFlow help me with this task or I am looking at this whole thing in a wrong way?