The use of nonlinear dynamical systems (such as compliant robot morphologies) as physical computing systems is widely spread in the domain of machine learning and robotics. Often, physical reservoir computing is used as a theoretical framework. In this project we will apply the physical reservoir computing to plants to quantify their internal computing power. This will allow us to better understand the dynamic responses of plants to their variable environment.
We will investigate the reaction of plants to environmental factors using imaging sensor technologies. A wide range of sensing devices will be used to capture the plant characteristics and vitals. This information will then be employed to characterize the nonlinear dynamical properties of the plant. Using these, we will conceive a general framework, based on the physical reservoir computing concept, able to analyze the computational processing capacity of plants. Ultimately, we aim to design a system in which the captured plant behaviour points out specific suboptimal environmental conditions.
The framework should offer a means to validate the information processing capabilities of plants. Furthermore, this will improve our knowledge on the complex plant responses to variable environmental factors like temperature, light and water availability. Ultimately, this should offer a means to improve management of crop growth e.g., climate monitoring, pathogen detection, greenhouse climate control and irrigation scheduling.
|Effective start/end date||1/10/17 → 30/09/21|