Where ever we use our senses. Eyes, smell or ?taste. An algorithm could perform ?better. So with the help of machine learned networks and python libraries or modules. I wonder if the you can feed the network start images of the target area and generate a time sequence of images. These then become the ”what to do” images.
The end image is then the finished greened environment.
So the idea is to build up and train weights for a Green Machine Learning Network. Like there exist other large networks.