- Home
- >
- Content Pages
- >
- Technical Resources
- >
- Pestology Blog Entries
- >
- ArTreeficial: An AI-tree Controlling Spotted Lanternfly Populations Using Computer Vision and Dynamic Response
ArTreeficial: An AI-tree Controlling Spotted Lanternfly Populations Using Computer Vision and Dynamic Response
The spotted lanternfly (SLF), Lycorma delicatula, has been a hot topic over the past few years. For those unfamiliar, this is an invasive species that has spread from its initial introduction around 2014 in PA to much of the eastern US. The spotted lanternfly is native to southwest China. This plant hopper is large, clumsy, and does not bite or sting. They use their piercing mouthparts to feed on the phloem of many plants including their favorite tree, the tree of heaven. This is an introduced tree from their native range that has been spreading rapidly and taking over resources and space from native plants. It came over in the early 1800s and escaped cultivation. The lanternflies feeding causes significant damage to the plants they feed on from the removal of phloem, and sooty mold that can grow from the honeydew that they excrete. There are concerns for grapevines, both wine and juice types. But in the context of urban and structural, they are a pest for their damage to trees and the sticky honeydew that grows sooty mold where it falls. Hundreds of lanternflies can be found on a single tree, excreting a rain of honeydew covering cars, people, and structures under them.
Current methods of controlling SLF numbers have not been widely successful due to the their high fecundity where a small base can grow into a large population quickly. There are some insecticidal products that show potential effectiveness but are non-specific and require repeated applications. The other option of using sticky trap bands around trees can get non-target animals stuck or get filled up with lanternflies. Even the public campaigns for folks to stomp on sight are only so effective.
To combat this problem, Selina took her knowledge of AI technology and combined it with as she states, “inspiration from chess and dance dance revolution” to create ArTreeFicial. This ai “tree” provides a low labor, cost efficient, and environmentally friendly solution. The tree uses tree of heaven based incense which has been shown as an attractant to SLFs. The incense is blended tree of heaven that has been strained. This is placed onto an outdoor umbrella which provides shade and protection from weather. The trunk of the umbrella/tree is wrapped in electricity conducting nets and cameras are set up to watch the trunk. Spotted lanternflies crawl up trees and are poor fliers so a band of netting around the trunk is an effective shape to catch most of them.
A computer model called YOLO identifies the SLF and selectively shocks that area of the trunk, killing the bug. The section of netting is two layered so when detected, the inner layer is activated and the bug on it completes the circuit, only killing it and nothing else. There is no live electricity running constantly. This is a similar to the electric fly swatters you may be familiar with. To round out this well designed product, there’s even a collection bin at the base to keep the dead ones.
The initial building cost of the tree is 150-200$ and Zhang predicts this could be lowered by mass manufacturing. This is a very accessible price which I could envision would lead to ease of use. Imagine a pest control company renting this out during the season to clients. Such a product could be widespread and profitable.
The article discusses that this type of device model could be used for other pests in the future with the attract and kill concept using computer vision for pest ID. The particular design works quite well considering the behavior of the SLFs. For occasional invaders or other structural pests, such devices would require some redesign, but it brings some hope for mechanical control options in pest control.
Article by: Ellie Lane
References
S. Zhang, "ArTreeficial: an AI-tree controlling spotted lanternfly populations using computer vision and dynamic response," 2023 IEEE International Conference on Advances in Data-Driven Analytics And Intelligent Systems (ADACIS), Marrakesh, Morocco, 2023, pp. 1-6, doi: 10.1109/ADACIS59737.2023.10424285.
keywords: {YOLO;Computer vision;Technological innovation;Sociology;Pesticides;Solar panels;Statistics;Spotted lanternfly;Lycorma delicatula;computer vision;agricultural application of computer vision;yolo;ailanthus altissima},
https://www.smithsonianmag.com/innovation/this-high-schooler-invented-an-ai-powered-trap-that-zaps-invasive-lanternflies-180983918/
Podcast