The broader impact of this Small Business Innovation Research (SBIR) Phase I project will be in protecting natural resources with actionable information on invasive insect pests to be used by local, state and federal agencies to monitor insect pests in real time and plan optimal intervention strategies. Invasive species disproportionately affect vulnerable communities in poor rural areas, especially in developing countries, who depend on natural resources, healthy ecosystems, trade and tourism for their livelihoods. Moreover, invasive insect pests can drive food insecurity and undermine ongoing investments in development. There is a growing consensus that "Early detection and rapid response" is the best solution, but the "Early detection" part of this equation has been lagging for want of robust automatic surveillance systems.The proposed project will investigate how to generalize existing machine learning models for insect classification under more general conditions. In particular, existing machine learning models typically make strong assumptions about the "priors" (the prior probably of seeing a given insect) due to the use of very specific attractants. However, to reduce the cost of deployment of insect surveillance, we will investigate the design of pheromone "cocktails" (combination of two or more pheromones) to attract multiple insect species to a single trap. This will require creating new models that do not make such strong assumptions about the insects to be encountered. In addition, the project will investigate novel design principles to create "compromise" traps. For example, if insect A is attracted to the blue color, and insect B is attracted to the red color, is the best color attractant a mixture of the colors (green), or a patchwork work of two colors be arranged as panels, tiles, stripes, etc.? Beyond color, we proposed to investigate the optimal compromises for texture, shape, trap placement, trap orientation etc. Thus, our proposed innovation will lie in producing sensors/physical traps that can simultaneously monitor multiple invasive pests.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
