Simulation and optimization of robotic tasks for UV treatment of diseases in horticulture

Abstract

Robotization is increasingly used in the agriculture since the last few decades. It is progressively replacing the human workforce that is deserting the agricultural sector, partly because of the harshness of its activities and health risks they may present. Moreover, robotization aims to improve efficiency and competitiveness of the agricultural sector. However, it leads to several research and development challenges regarding robots supervision, control and optimization. This paper presents a simulation and optimization approach for the optimization of robotized treatment tasks using type-c ultraviolet radiation in horticulture. The optimization of tasks scheduling problem is formalized and a heuristic and a genetic algorithms are proposed to solve it. These algorithms are evaluated compared to an exact method using a multiagent-based simulation approach. The simulator takes into account the evolution of the disease during time and simulates the execution of treatment tasks by the robot.

Phenotypic Variation of Botrytis cinerea Isolates Is Influenced by Spectral Light Quality

“…Today robots equipped with sensors are under investigation in greenhouse production. A UV-Robot with UV-C radiation is developed to control powdery mildew in horticulture (Mazar et al, 2018). This could also be equipped with imaging sensors, as early detection of disease is beneficial from both economic and environmental perspectives.…”

Smart greenhouses as the path towards precision agriculture in the food-energy and water nexus: case study of Qatar

“…Another improved robot, comprising ground and aerial vehicles, has been proposed to measure the temperature, humidity, luminosity, and carbon dioxide concentration in the ground and at different heights of the greenhouse (Roldán et al 2016). However, robotic systems have not been used only for environmental monitoring, but also to perform greenhouse unsafe tasks like the UV-C treatment against powdery mildew (Mazar et al 2018). In fact, robotic systems have been designed to help labor tasks such as pesticide spraying with path-planning and traveling-control abilities (Nakao et al 2017;Mahmud et al 2019), transplanting (Han et al 2018), fertilizing (Yuan et al 2016), phenotyping (Atefi et al 2019) and harvesting (Wang et al 2017b;Arad et al 2020;Woo et al 2020).…”

Multi-agent Simulation to Predict Global Behavior of Population Based on Elementary Local Markovian model

“…The supervisor is able to collect data about the state of the plants through sensors in the greenhouses or drones in open-field farms. After collecting the data and extracting the information it contains, the supervisor decides whether to harvest the plants or treat them [3]. In [4], harvesting is ensured by harvesting robots, which can be customized regarding the type of plants and can work in group.…”

“…The manual deployment of this new technology cannot be possible due to the numerous and real risks such as burns or cancer. Robots equipped with UV-lamps are suitable for this kind of operations [3]. These robots have to move next to plant with turned on UV-lamps and to make some decisions related to the selection of row to treat, charging cycle planing and speed.…”

“…A multi-agent base simulator is then developed using NetLogo software to represent the behaviour of greenhouse. The greenhouse is a part of global simulation of UV-C treatment already presented in other works such as [3] and [12]. Among the existing agents of the model, the agent class " Plants " is considered as a holon [13], which contains several sub-agent " Plant ".…”