[Paper Review] Multi-UAV-based Optimal Crop-dusting of Anomalously Diffusing Infestation of Crops

Jianxiong Cao, YangQuan Chen+1 more|arXiv (Cornell University)|Nov 7, 2014

Plant Parasitism and Resistance8 references4 citations

TL;DR

This paper proposes a multi-UAV system for optimal crop-dusting of crops infested by pests exhibiting anomalous diffusion, modeled via time- and space-fractional diffusion equations. Using Centroidal Voronoi Tessellations (CVT) for dynamic actuator placement and a novel simulation platform (FO-DiffMAS-2D), the method minimizes pesticide use and environmental impact while effectively controlling pest spread, with fractional order parameters significantly influencing control performance.

ABSTRACT

This paper presents a UAV-based optimal crop-dusting method to control anomalously diffusing infestation of crops. Two anomalous diffusion models are considered, which are, respectively, time-fractional order diffusion equation and space-fractional order diffusion equation. Our problem formulation is motivated by real-time pest management by using networked unmanned cropdusters where the pest spreading is modeled as a fractional diffusion equation. We attempt to solve the optimal dynamic location of actuators by using Centroidal Voronoi Tessellations. A new simulation platform (FO-DiffMAS-2D) for measurement scheduling and controls in fractional order distributed parameter systems is also introduced in this paper. Simulation results are presented to show the effectiveness of the proposed method as well as the role of fractional order in the overall control performance.

Motivation & Objective

• To address the challenge of controlling pest infestations that exhibit anomalous diffusion, which standard models fail to capture.
• To minimize environmental and soil impact from pesticide application by optimizing UAV actuator placement and dynamic spraying.
• To develop a real-time, UAV-based control system for dynamic pest management in agricultural settings.
• To evaluate the influence of fractional order parameters (α and β) on control effectiveness in anomalous diffusion processes.
• To validate the proposed method through a new simulation platform tailored for fractional-order distributed parameter systems.

Proposed method

• Modeling pest infestation dynamics using time-fractional order diffusion equations (Caputo derivative) and space-fractional order diffusion equations.
• Formulating the control problem as an optimal actuator placement task in a distributed parameter system governed by fractional PDEs.
• Applying Centroidal Voronoi Tessellations (CVT) to compute dynamic, optimal UAV positions for pesticide release without relying on a full system model.
• Developing a new simulation platform, FO-DiffMAS-2D, to enable measurement scheduling and control in fractional-order distributed systems with mobile sensors and actuators.
• Implementing a control law that uses real-time sensor data (measured pest density) to guide UAVs in spraying pesticides at optimal locations.
• Validating the method through numerical simulations with exact solutions for both time- and space-fractional diffusion equations.

Experimental results

Research questions

• RQ1 How does the use of fractional-order diffusion models improve the representation of anomalous pest spread compared to classical diffusion models?
• RQ2 What is the optimal dynamic placement strategy for multiple UAVs to minimize pesticide use and environmental impact in crop-dusting applications?
• RQ3 How do the fractional order parameters α (time) and β (space) affect the controllability and performance of the UAV-based crop-dusting system?
• RQ4 To what extent does the CVT-based actuator placement algorithm outperform non-optimized or static placement strategies in controlling anomalous diffusion?
• RQ5 How effective is the FO-DiffMAS-2D simulation platform in replicating and validating control strategies for fractional-order distributed systems?

Key findings

• The proposed CVT-based UAV placement method effectively controls pest infestation in both time- and space-fractional diffusion models, significantly reducing the spread of pests.
• Numerical simulations show excellent agreement between the FO-DiffMAS-2D platform's results and exact solutions for both time-fractional (α = 0.6 to 0.9) and space-fractional (β = 1.3 to 1.9) diffusion equations.
• The optimal fractional order for control performance was found to be α ≈ 0.8 for time-fractional diffusion and β ≈ 1.7 for space-fractional diffusion, balancing response speed and stability.
• The simulation results confirm that the fractional order has a measurable and significant impact on control performance, with higher β values (e.g., 1.7) yielding better suppression of pest spread.
• The FO-DiffMAS-2D platform successfully simulates and validates the control strategy, demonstrating its capability to handle complex, real-time measurement and control tasks in fractional-order systems.
• The method achieves minimal environmental impact by concentrating pesticide application at optimal, dynamically updated UAV positions, reducing overall chemical usage.