Project
Context
The use of pesticides appears natural and exclusive to us because our civilization has relied on them since antiquity. Indeed, pest damage results in economic production losses to the agricultural industry, estimated from 28 to 50% (in Africa and Asia) of annual productions. Therefore, the European Union uses approximately 360 million kg of pesticides per year for agricultural and horticultural tasks. However pesticide application methods are inefficient (only 0.3% of sprayed pesticides from aerial application comes in contact with the target pest) and has led to alarming consequences in public health, the environment, and economically. In 2003, environmental and economic costs associated with pesticide use were estimated to total approximately 10 billion dollars per year in the US. In its “Ecophyto” plan, French government has decided to reduce by 50% the use of pesticides by 2018. Unfortunately, alternatives to pesticides being too scarce, the objective has been postponed to 2025. So far, no purely technological and versatile method has been developed to replace pesticides. Detection techniques still do not make use of spectrometry to detect pests (they detect sick plants, so too late). For instance, in 2012, some robots have collaborated with humans in vineyards to decide where to spray with various levels of autonomy. Results showed 90%1accuracy of grape cluster detection leading to 30% reduction in the use of pesticides. Some studies introduce ways to detect pests on leaves with a camera, but due to the challenges of on-site detection, most of them relied on scanning under highly controlled light conditions.
1 Ron Berenstein, Yael Edan, Human-Robot Cooperative Precision Spraying: Collaboration Levels and Optimization Function, 10th IFAC Symposium on Robot Control, September 5-7, 2012. Dubrovnik, Croatia
Objectives
Greenshield Project aims at reducing the use of pesticides by developing a robotic module to be embedded on a terrestrial vehicle (mobile robot, farming tractor, ...) to fight against crop pests (invertebrates, diseases, weeds). This module will autonomously detect pests using spectral analysis, and destroy them with a laser. When mounted on mobile robots, it will patrol among crop fields to scan the plants, collect accurate data regarding pests that will serve to optimize the action of robots. This new means of fighting will settle a new sustainable paradigm of pest control to better combat them.
Pre-Existing results
The technical solutions proposed in this project have already been proved to be feasible. It is established in the scientific litterature that, by way of spectrometry, one can characterize insect species with more than 95% of correct answers, and even 80% of correct answers for subspecies identification. In January 2017, we performed and succeeded a Proof Of Concept (P.O.C.) study validating this method of detection on aphids with a spectrometer off-the-shelf. It was performed with non optimized tools (simple statistical method, spectrometer off the shelf). Concerning the destruction method, its effectiveness has been successfully studied on cockroaches. The methods for detection and destruction envisaged in this project have been previously patented by the Green Shield Technology start-up, which will use these project results to industrialize the robot, and thus, will answer to a common need expressed by firms of wine market for instance.