Click on “Download PDF” for the PDF version or on the title for the HTML version.
If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.
Real-Time 3D Tracking of Flying Moths Using Stereo Vision for Laser Pest Control
Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org
Citation: 2023 ASABE Annual International Meeting 2300079.(doi:10.13031/aim.202300079)
Authors: Ryo Sugiura, Ryo Nakano, Kazuki Shibuya, Koji Nishisue, Shinji Fukuda
Keywords: stereo vision, image recognition, Kalman filter, pest control.
Abstract. Pest management is a critical aspect of crop production, and the current approach to controlling pests involves using chemical pesticides. However, the long-term use of chemicals has led to an increased occurrence of insecticide resistance, as well as environmental concerns. Integrated Pest Management (IPM) has been identified as a promising solution to mitigate the negative impact of chemical use and reduce crop losses associated with pests. This study presents the first step of a program aimed at developing a novel physical pest control method in the form of a laser zapping system to eliminate flying pests in fields as part of an IPM strategy. This study developed a method to track the three-dimensional (3D) position of flying pests in real time. Stereo image pairs were captured at 55 frames per second (fps) while moths (Spodoptera litura) were flying, and time series 3D point cloud data were generated from the images. Point cloud filtering was then developed to successfully extract the point data of the individual moths while removing the other unused points (such as walls or noise in the air). The proposed filtering method focused primarily on the size and motion of the point cluster. Machine learning techniques were used to identify the point cluster as a moth. Moreover, a Kalman filter was applied to predict the position of the moth to be zapped using a laser beam. All these methods were incorporated into the software to simulate predicting the position of flying moths and zap them with a laser. The simulation experiment results revealed that the position of a moth could be predicted with a root mean square error (RMSE) of 1.4 cm. Future work on this program will involve prototyping the system by integrating stereo vision and a computer to process images and control galvano mirrors to adjust the direction of the laser beam.
(Download PDF) (Export to EndNotes)