Progress 07/01/19 to 09/30/20

Outputs
Target Audience: Nothing Reported Changes/Problems:The first and most obvious of the major changes and problems was COVID, which significantly slowed the completion of this research. The research was awarded later than expected and conflicted with other existing projects. We applied for and got an extension, which unfortunately pushed this project into the core COVID shutdown period. Regardless, we were able to complete the project. Our original research objective was to compare machine vision to human vision and determine if the machine imagery was capable of detecting nozzle faults at at least a similar rate to human vision. This proved to be a relatively low bar, as it has been shown that humans cannot see faults until there is thirty to 40% mis-application byt the nozzle or worse. Upon investigation of the thermal imagery apparati it immediately became clear that comparison to human perception was folly; the sensitivity of the imagery was far superior, and our approach shifted to an intense understanding of the image itself and computational methods for comparison of operational variables between nozzle operating parameters and atmospheric conditions. The results of this are currently proprietary and will be reported in detail in the Project Final Report, The Phase 2 application and the next Phase I application. As part of validating the sensitivity of the imagery, we found that within the North American Continent, there were no analog device that could measure with sufficient precision the distribution of spray nozzles to the level that would allow us to truly and precisely validate the machine vision. So we developed and validated a device to do that, and to our knowledge it is the only spray patternator in North America that is capable of automatically recording spray distribution from a nozzle or array of nozzles at 25 mm resolution and to an r2 value of .99 or better. This is a benefit to American industry as a whole as it will remain available as a research tool on a contract basis within our company to anyone that would need such a tool. An unexpected outcome was the level of precision we were able to draw from the data. This caused us to invest significant additional time and resources to follow that trend. It is resulting in the pursuit of a second SBIR Phase I proposal. What opportunities for training and professional development has the project provided? Nothing Reported How have the results been disseminated to communities of interest?An Inter-Agency Agreement with Michigan State University and Application Insight was developed and agreed to in order to foster the continued development of this technology and fully leverage the resources of both entities. MSU will control and ultimately be responsible for the licensure of this technology moving forward. The Intellectual property has not been submitted for patent consideration so it has not been further disseminated. Scientific papers are expected to be presented at relevant symposia in late 2021 or early 2022. What do you plan to do during the next reporting period to accomplish the goals? Nothing Reported

Impacts
What was accomplished under these goals? Project Impact: This SBIR overwhelmingly achieved its technical objective of proving that digital thermal imagery can be used to reliably and repeatably detect spray nozzle faults in a wide range of scenarios consistent with those found on agricultural airblast sprayers. Indeed, we discovered that there was more information that we originally anticipated in the data stream and not only are we able to continue to SBIR phase II with the technology, but we also intend to submit a second Phase I application to explore additional more precise applications of the core technology. The current embodiment of the technology will require significantly more thermal imaging devices per sprayer than originally conceived, and the placement of the imaging devices will be more complicated. This would have made the project useless in the original target of all orchard sprayers as it would not be economically viable. Serendipitously, advances in other technology since the original proposal have opened new opportunities within that sector that are smaller markets in terms of number of total units, but at higher price points and higher likelihood of adoption; these markets could total $30 million per year for this technology by 2023, with a 10 year potential of $300,000,000 in total market for this technology. Major Activities Completed/ Experiments conducted Our original research objective was to compare machine vision to human vision and determine if the machine imagery was capable of detecting nozzle faults at at least a similar rate to human vision. This proved to be a relatively low bar, as it has been shown that humans cannot see faults until there is thirty to 40% mis-application byt the nozzle or worse. Upon investigation of the thermal imagery apparati it immediately became clear that comparison to human perception was folly; the sensitivity of the imagery was far superior, and our approach shifted to an intense understanding of the image itself and computational methods for comparison of operational variables between nozzle operating parameters and atmospheric conditions. The results of this are currently proprietary and will be reported in detail in the Project Final Report, The Phase 2 application and the next Phase I application. As part of validating the sensitivity of the imagery, we found that within the North American Continent, there were no analog device that could measure with sufficient precision the distribution of spray nozzles to the level that would allow us to truly and precisely validate the machine vision. So we developed and validated a device to do that, and to our knowledge it is the only spray patternator in North America that is capable of automatically recording spray distribution from a nozzle or array of nozzles at 25 mm resolution and to an r2 value of .99 or better. This is a benefit to American industry as a whole as it will remain available as a research tool on a contract basis within our company to anyone that would need such a tool. Data collected: The Data collected is currently proprietary but it is representative of the range of application scenarios that would be seen in orchard spraying scenarios, not only in nozzle type and flow, but in typical damage scenarios. Another method we developed as part of this project was a method to reliably and repeatedly wear spray nozzles in order to create reference sets. Again, it is anticipated this will become a useful product for test labs all over the world. Summary Statistics and Discussion of results: Intellectual Property protection will be pursued for this project as a fair degree of novel insight was produced including discoveries that will lead to an additional SBIR Phase 1 application; therefore the results cannot be discussed in detail here. Key outcomes and other accomplishments realized: The capability of the thermal imagery in this context was understood sufficiently to justify advancement of the project to Phase II Although the implementation and embodiment of the technology changed to a more expensive configuration, new agricultural markets that will be receptive to the technology and that will profit from it were identified. This has the potential to create $300 million in additional economic activity over 10 years along with other significant economic benefits that have not otherwise been identified. New methods for patternation of nozzles at higher resolution and with automated, highly statistically relevant data collection that far exceeds manual collection, which was not previously available to the agricultural community in North America, were developed. Methods for accelerating the wear of nozzles to create reference sets of worn test nozzles were refined and codified. A path to a second Phase I project using the thermal imagery for another important process in Agricultural chemical application was defined. As the original genesis of this project was with Dr. Matthew Grieshop at Michigan State University, an Inter-Agency Agreement with Michigan State University and Application Insight was developed and agreed to in order to foster the continued development of this technology and fully leverage the resources of both entities. MSU will control and ultimately be responsible for the licensure of this technology moving forward. This allows Application insight to "punch above its weight" so to speak with the resources of a large research university managing the IP.

Publications