So I did not see a topic for this already so I would go ahead and start one because I have ideas I would like to throw out.
Human interaction is still going to be necessary for farmbot for some time. Some software bits could greatly enhance those interactions that must happen.
My main request is a microphone and small speaker on the control housing. While your hands are dirty from working in the garden for harvest for example, use voice recognition to command the robot. The speaker gives command recognition confirmation, and can play music while you dilly around in the dirt.
So for example, in polycrop different plants will be ready at different times. Farmbot gives you a notification through its phone app that some plants are ready or need human tending. Then when you go out to harvest, the bot “points” to where the mature plant is. You can then dig it up, and while hands still dirty, voice command “next finished plant” or something and it jogs and points to the next. This could also be used for plants that it expects need to be pruned or trimmed for various reasons such as flowering starting in plants that you do not want to flower. It can “point” to everything that needs human interaction and without you touching anything.
You could also integrate a weight scale, and weigh every plant you harvest (and the subroutine will be pointing to that plant as you harvest it) so you can get yield data based on some experimental watering/scheduling/other variables you might be performing. Maybe you are playing around with how much water to give to a particular species. This would give you the ability to quantify your experiment in a catalog-able way.
One of the more critical factors to making this system really great is having good data behind everything. I see OpenFarm is being intergrated, and this is extremely important. Good data on each plant, growing space, synergies, nutrient needs, waste products, all that stuff. It would be kind of cool to have a “simulator” kind of like the game of life that takes in the plant attributes and runs through several life cycles of adding and subtracting nutrients from the soil, insects, random factors, harvests and optimizing the farm layouts automatically based on the best yields from these models. It is outside the scope of where the project is now and is really a whole project in its own, but it would be super cool.