AI-controlled dynamic root competition for soil resources
Martin Bouda, Štěpán Kříž
Plant growth requires soil resources, whose distribution can be strongly variable in both space and time, with resource-rich patches and resource-poor gaps occurring down to scales of centimetres and hours. It remains unknown how plants can best explore and exploit this heterogeneous resource domain. We are presently building a custom artificial neural network to analyse root-system structures and eventually control their growth in dynamic simulations of resource exploitation. We aim to investigate the spatial and dynamic aspects of root growth and competition, addressing previously intractable questions. This work will show how ecological interactions among species arise from root growth and uptake processes. More broadly, as an exploration of Artificial Intelligence optimising networks that compete in resource acquisition and transport, we expect this work to find applications beyond plant ecology.