COS 8-3
Feedback control-based design for robust pest management

Monday, August 10, 2015: 2:10 PM
322, Baltimore Convention Center
Stuart Townley, Environment and Sustainability Institute, University of Exeter, Penryn, England
Chris Guiver, Environment and Sustainability Institute, University of Exeter, Penryn, England

The 21st century is likely to see increasing global demand for food owing to a growing population, significant increases in per capita consumption and climate change. A key challenge to meeting this increase is to reduce losses in crop yield to weeds and animal pests, while minimizing the environmental impact of agriculture. The prevalent method for managing pest populations is the use of chemical pesticides. Worldwide use of pesticides is huge - farmers applied approximately 2.5 billion kg of pesticides in 2007. With huge social and economic drivers to securing food and reducing crop loss, there are significant resources devoted to managing effective application of pesticide. The prevailing approach is to base management strategies on optimization techniques predicated on minimizing cost or maximizing production. But these approaches often lack sufficient robustness to cope with the uncertainties that arise in modelling population dynamics of pests.


Drawing on ideas from robust control engineering, we propose a novel approach to robust pest management based on adaptive feedback control.  In the context of pesticide application, the management strategy `adapts' the mass of pesticide to be applied as a function of measured pest abundance in a feedback arrangement. The strategy can be likened to adaptive management except that here the adaptive feedback control does not attempt to infer or update system parameters. Instead, the control design is guided by structural properties of the pest dynamics. Efficacy of the strategy is not reliant on detailed knowledge of the pest dynamics, an appealing feature given the uncertainties present in modelling pests, but does utilize monotonicity properties inherent of population dynamics.