Recipient Organization
GEARJUMP TECHNOLOGIES LLC
216 SUMMIT AVE UNIT E202
BROOKLINE,MA 024462331
Performing Department
Engineering
Non Technical Summary
The Red Imported Fire Ants (RIFA) infest over 140 million hectares in the USA and cost Americans an estimated 7 billion dollars annually for control and to repair damage to a number of economic sectors, including Agriculture, Nurseries, Sod Producers. It is estimated that at least 30% of the population in RIFA infested areas is stung per year. Overall RIFA infestation has resulted in higher food production costs, increased use of pesticides (that also harm other beneficial insects), increased medical and veterinary costs, significant damage to equipment and ecological impact.The Remote Autonomous Detection System (RADS) represents a novel surveillance platform designed to detect RIFA cost effectively. RADS relies on integration of electronics and data analytics to detect, identify such an invasive species.RADS relies on the integration of microfluidics, electronics and special digital processing algorithms to accurately detect RIFA. The system will operate autonomously for 4-8 weeks and will be characterized by its small form factor (15 cm x 5 cm x 10 cm) and ease of operation.The main objectives for Phase I are: to demonstrate a proof-of-concept that will attract, capture and detect RIFA, as well as positively identify it with a high degree of certainty using custom design digital algorithms. Entomological experiments will be performed at the USDA ARS facilities to demonstrate the concept. The main objectives for Phase II is to demonstrate a fully functional product-grade prototype in large entomological trials, consolidate Intellectual Property (IP), and develop a business plan for mass production and commercialization.Commercial applications of the RADS platform include a number of sectors, including surveillance in Ports, Ships and Agriculture Fields. The target cost of this device is below $100, which will allow massive deployment. RADS represents the next generation surveillance tools to protect the US territory against RIFA and other invasive species, providing an effective preventive tool for the US agricultural industry. Overall, the deployment of RADS will allow significant improvement in the environment, as well as savings in multiple economic sectors and overall increase production US agricultural output.
Animal Health Component
30%
Research Effort Categories
Basic
10%
Applied
30%
Developmental
60%
Goals / Objectives
The main objectives of Phase I will include the basic development of a RADS that will integrate the following:Controlled Release System. Controlled release system that is able to spatially deliver a pheromone to increase attraction of RIFA.Trap and Camera System. A trap that integrates a micro-controller with camera system that is able to capture and detect the presence of insects;Image Identification. A digital processing system that is able to integrate machine learning algorithms to positively identify RIFA among other insects.
Project Methods
The following tasks summarize the methods proposed for this project.Task 1: Formulation Selection and Physical Characterization. The formulation and characterization of the attractants is the first step in the development of the controlled release system. This task therefore involves characterization of evaporation rates of the formulated pheromones by multiple analytical methods, e.g. Thermo-Gravimetric Analysis (TGA) or GCMS (Gas Chromatography and Mass Spectroscopy), to characterize the evaporation as a function of time. These rates are key for determining the targeted specific 'bubble' or volume of attraction. Formulation of the attractants will be performed using combination of natural oils and organic solvents in order to decrease the evaporation rates. Initial candidates for organic solvents are petrolatum and paraffin oil, which are inert mineral oils.Task 2: Basic Controlled Release System. The physical characterization is completed and optimal formulation is selected, a controlled release system will be designed and implemented to deliver the attractant for targeted duration for a given space volume. Initial targeted specifications will require a 4-6 week (depending on volume size) duration for a volume of ~ 6 m x 2.5 m x 2.6 m, e.g. ship container volume. Evaporation rates will be used for numerical analysis based on CFD simulations to obtain spatial dispersion of attractants over time.Attraction tests with RIFA will be performed to characterize attraction efficacy in a restricted volume. Concentration vs attraction correlations will be established and used to tune the controlled release system and the reservoir volume to satisfy an attraction requirement. The overall goal of this experiments will be to determine the sensitivity of the system, as well as the coverage area for indoor and outdoor applications. Once controlled release rates of the RADS are optimized, it will be possible to characterize the 'sensitivity' obtained as a ratio of ants detected per coverage area for a given height of the indoor facility (volume) for a given temperature. It is important to consider that for indoor applications the height plays a role as it defines the overall volumetric space in which the controlled release system will operate. This metric will be critical. Additionally, we will evaluate the 'specificity' of the detection system by performing multiple controls with and without pheromones with RIFA and other non-targeted species. This evaluation will allow to obtain a quantitative metric on the specificity of the system.It is important to create a concentration gradient to lure RIFA into the RADS. If the specific coverage volume is saturated with the pheromone, then the ants will be confounded or habituated and will not be directionally attracted to the RADS. A critical part of this task is to determine the performance metrics of the RADS by quantifying detection with and without the use of a pheromone.Task 3: Design and Manufacturing of Housing.The design and manufacturing of a basic housing system that include compartments for integration with the controlled release system (1); Electronics module (2); large external battery for prolonged performance (3). Housing will be designed using CAD software. The key design feature will provide access to ants from all directions to the arena area guided by the source of the attractant. The design will also include the sticky trap as part of a cartridge module designed to capture and retain RIFA under the camera field view. Housing of the device will be manufactured via 3D printing for rapid and precise prototyping. The standard sticky trap and customized controlled release (pheromone) module will be part of a replaceable cartridge that could be quickly swapped.?Task 4: Image Acquisition System and Electronics Integration. The implementation of standard electronics module based on a micro-raspberry pi micro-controller, which will integrate a high resolution camera for image acquisition. The micro-controller can be easily programmed to adjust acquisition features, including sampling rate, image resolution and data uploading. The RADS will be able to operate on two modes. The main mode is long autonomy, which requires low power consumption. For this mode, 1-24 images will be taken a day to conserve power for weeks to months.Task 5: Image Identification Algorithms Image identification algorithms will be implemented and custom developed from available libraries, including segmentation and cross-correlation based pattern recognition algorithms, to identify the RIFA based on distinctive features that include: color, head and antennae, legs, body and overall proportion. These advanced algorithms are readily available in open source for a number of image recognition applications. Multiple images of RIFA will be used as a control to compare them with captured images. Machine learning algorithms rely on multiple iterations to increase identification accuracy, which also includes orientation. In addition, spectral compositions of images will allow to correlate the color signature for RIFA. This task will therefore include characterization of automated recognition system that will result in a set of criteria for matching separate features and provide an overall score. This score will include quantifying detection. Computer vision algorithms will be based on multuple types of technqiues, e.g. Convolutional Neural Networks (CNNs) with foreground extraction, feature recognition and contour identification. CNNs are primarily used for image classification and recognition because of its high accuracy. Additionally, Data Augmentation algorithms could inlcude crop augmentation to enlarge datasets. Background Removal algorithms could also be used to remove subjects already classified. These known algorithms allow to use actual available datasets for both RIFA and other insects. Datasets on specific features related to multiple ant species as well as other non-targeted insects are available from USDA, antweb.org, as well as other public resources. Antweb.org contains one of the largest databases of image records of multiple ant species.Overall, several techniqueswill be explored, selected, optimized and used for image analysis to implement custom identification algorithms including colors and features.Task 6: In Vivo Testing and Characterization with RIFA. This task involves iterative characterization of the controlled release system in a lab setting initially to test sensitivity, specificity and overall identification process. Experiments will be performed at Dr. Vander Meer's USDA laboratory and will consist of testing the system with fire ants in indoor and outdoor applications. For indoor applications, the system will be deployed inside of a container, volume approximately 10 m x 2 m x 2 m (40 m3). The system will rely on the natural foraging efficiency of the ant to find and enter the RADS.The goal of the initial experiments is to test the system in a closed environment similar to a ship container where environmental variables can be controlled. The objective of this experiment is to test the sensitivity of the system for a given volume. If time allows during Phase I, we will add preliminary tests outdoors (otherwise, it will be deferred for Phase II). In this case, the RADS will be tested in an outdoor environment to test the system performance in terms of sensitivity in outdoor environmental conditions, e.g. wind, rain, temperature and humidity variations. In all experimental evaluations, if non-targeted insects are trapped, they will be noted and characterized.