This thesis introduces novel technological tools related to insects from various angles: hardware platforms that detect and register insect’s presence, data processing techniques and machine learning classification algorithms that identify the species. But why there is such necessity to detect and classify the humble and sometimes, unnoticed insects? Some insect species (moths and butterflies for example) are considered suitable indicators for biodiversity that is currently declining globally at an unknown and unmeasured rate. All insects play a vital role as they are discrete links in the biological dependence of species and are food sources for vertebrates. Therefore, they are not only important, but they are also crucial. However, many of them can reach high, damaging densities and severely impair plant processes in agricultural (several flies like the medfly, dacus, bactrocera dorsalis) and forest (e.g., wood boring insects) ecosystems. Some other like mosquitoes and biting midges pose a serious hygienic threat making the mosquitoes the deadliest animal on the planet. However, If we would prioritize the significance of insects to humans, we would describe this priority with a single phrase ‘they impact food production’. A grand challenge facing the 21st century is the sustainable production of food for the growing human population. The world population is projected to reach 10.4 billion people sometime in the 2080s. Technological means with the increasing availability of new sensors and electronic components and the associated collection of ‘big data’ aim at improving the sustainability of food production systems. The urgent necessity is to moderate crop losses due to insects while avoiding unnecessary spraying that has been deemed to impact human health in a negative way. Agronomists had always deployed traps (usually bottles or buckets) to trap and count pests as a means to assess the infestation situation. With time they had developed the so called ‘decision protocols’ to follow as a rule of thumb of what to do in the presence of a specific density of pests in agricultural fields and forests. In practice, reports from monitoring traps are not accepted blindly but serve as supportive evidence. Certified state entomologists adapt the rule to the particularities of different geographical parts of the country and integrate diverge sources of information before granting permission for large-scale spraying. Reports coming out of manual monitoring of traps are accepted with a varying degree of trust. The main difficulty of this procedure is that a large number of traps must be strategically placed in orchards, sometimes on distant and remote locations and numerous people should place, maintain and inspect the traps on a 5-day basis from the end of spring till the end of fall. The pest-managers must discern the pest in the mass of collected maze of dead insects and even extract and deliver the pest to authorities for verification. This procedure is complicated, it involves a large number of people that are not always qualified to carry the task but, most importantly, can be easily bypassed by practitioners. Therefore, large economic loss is often reported because of the pest and this is usually attributed by expert entomologists not to the inefficiency of the monitoring protocol but to its opportunistic application that often leads to an ‘educated guess’ of when and where to start the treatment procedure. From both a conceptual and management perspective, there is an urgent need to increase the information flow over large areas and through time from the field-traps and in other cases probes, straight to a central monitoring agency, as well as to visualize and summarize this flow in a statistically reliable sense. To this end, this thesis develops technologies to improve, expand and automate global monitoring of insects of economic and hygienic importance. Τhis study required intensive interdisciplinary collaboration and consultation with specialists and experts from various disparate fields of science such as: low-power electronics, optoelectronics, data science, artificial intelligence, and entomology. Automatic detection of such insects is crucial to the future of crop protection by providing critical information to the appropriate personnel in a timely manner to assess the risk to a crop or a tree and the need for preventative measures. We develop novel single and multispectral sensors (hardware and microprocessor’s software) to record the wingbeat of flying insects. The wingbeat frequency and its harmonics are associated to species identity like the human voice to an individual. We describe new optical sensing techniques in which the light scattered by the beating wings of a flying insect and its casted shadow variations are recorded by a sensor that outputs a recording. We fabricate novel bee-counters that are attached to beehives and count the traffic of incoming/outgoing bees. We subsequently developed a new type of sensor that is attached to trees and records the woodboring insects the chew or move inside the tree and wirelessly uploads the vibrational snippets to a remote server. Finally, we introduce the electronic funnel trap (e-funnel) of automatic monitoring for all the Lepidoptera species with known pheromone. The e-funnels carry an optical counter that counts captures of Lepidopterans and form their own network based on the long range (LoRa) radio technology. The gateway of the network reports collection results of insect counts, GPS locations, timestamps and temperature to a cloud server.
.jpg){kind=logo}
University of West Attica
