Photonics Spectra spoke with Kiri Miyaca Fløistrup, an agronomist and science writer for FaunaPhotonics ApS, about a potential solution to reduce insecticide use in conventional farming. Fløistrup has a master’s degree in agriculture from the University of Copenhagen. What was the motivation that inspired insect monitoring? In the early 2010s, researchers Mikkel Brydegaard Sørensen at Lund Laser Centre and Carsten Kirkeby at DTU VET (Technical University of Denmark, National Veterinary Institute) were working on ways to combine research in lidar technology with entomology. The idea stemmed from papers published in 2005 to 2007 by a group led by professor Joseph Shaw from Montana State University that used lidar to detect bees. Initial research was done monitoring damselflies with atmospheric lidar, but the system was clumsy and heavy and could only count around 500 damselflies per week. After finishing his Ph.D. in 2012, Brydegaard Sørensen developed Scheimpflug lidar. (He won the Inaba Prize for this in 2015.) This system was much easier to transport because it only weighed around 100 kg. It measured with 10,000 Hz and could detect around 500,000 insects per week. Frederik Taarnhøj, Brydegaard Sørensen, and Carsten Kirkeby co-founded FaunaPhotonics in 2014, initially to develop sensors for monitoring different aerial and aquatic fauna. A copy of the Scheimpflug lidar system was the first sensor used in the company, but we have come a long way since then. Through the years we have become more focused on monitoring insects and zooplankton. FaunaPhotonics is now a growing company operating internationally. Monitoring insects is a labor-intensive, time-consuming, and costly activity, but finding an automated solution is expected to significantly improve knowledge and decrease costs. At the same time, insect pests directly damage crops, causing significant losses. Efficient pest control that is nonthreatening to beneficials and pollinators is a major challenge. The ability to systematically, efficiently, and accurately monitor insect populations is key to improved pest control. Courtesy of FaunaPhotonics. Please describe your technology, how it functions, and what data is collected. Our sensor technology has really progressed, and we are currently working with our fifth-generation sensor. We are further away from the initial lidar-based technology, though, and use much cheaper and more flexible LED lights instead. An LED module illuminates an air volume in front of the instrument with two overlapping conical beams at 810 and 970 nm. Much like the way you see light reflected off moths in car headlights at night, insects that fly through the volume reflect the emitted light back toward the sensor. The reflected light from any insect passing through the detection volume is collected by a silicone quadrant detector. Both wavelengths are collected on the same detector and separated using lock-in amplification implemented in an FPGA (field-programmable gate array) module. The resulting raw data consists of eight time series sampled at 20 kHz, from which insect events are automatically extracted by the sensor and transmitted via GSM (Global System for Mobile Communications) to our cloud for storage and processing. The detection volume is determined by the overlap of the illuminated volume and the detector field of view, and ranges between 30 and 150 cm (Figure 1). Figure 1. FaunaPhotonics’ insect sensor. LED module (1), illuminated volume (2), detector field of view (3), photodiode quadrant (4), detection volume (5), and antenna (6). Various features can be extracted from the recorded events. Most important is the insect’s wingbeat frequency, which is clearly visible as a modulation on the recorded signal. Other features are wing-to-body ratio and color. The color is determined by the amount of melanin in the insect, which is estimated by the difference in absorption at the two wavelengths. The automated transfer of data directly to the cloud ensures real-time monitoring of insect activity in the field with a day-to-day, hour-to-hour, or even minute-to-minute resolution. Using machine learning networks, the data is processed to determine insect species or more general insect activity during the day. The algorithm is trained on data collected in laboratory and close to field conditions on reared and captured insects. As the network gradually improves, we will be able to provide more and more information on the insect distribution in the field. This knowledge gives farmers insight into field conditions and can give warning if pests arrive or if activity from a certain pest is going up or down. It also means the farmer has better tools to decide when and where to spray. Our goal is a 20% reduction in the use of insecticides in conventional farming when using the sensor. As a business, at what stage is FaunaPhotonics, and what are your near-term plans? We are at the point where we will begin selling our product to customers in 2020. A lot of different projects will help us in this journey throughout the year. We will be working on monitoring pollen beetles in winter oilseed rape in Denmark and France, and pest insects in other crops with a collaboration partner in the U.S. Biodiversity will be a larger part of our company going forth, and we plan to work with partners in Germany, France, Ireland, and the U.S.