Progress 01/01/21 to 06/30/23
Outputs
Target Audience:The target audience reached during the reporting period was: Graduate and undergraduate students: Directly involved over three dozenuniversity students (including female minority students) in seven disciplines - electrical engineering, materials science and engineering, computer science, management of technology, industrial design, business information systems, and agribusiness. In addition to the PhD, MS, and undergraduate students supported under the grant, we also reached two Electrical Engineering capstone teams (four members each), a Computer Science capstone team (five members), two teams (comprising 10 students in total) in the multi-disciplinary "InnovationSpace" capstone program (organized by the Design School), a senior project team in the Center for Entrepreneurship in the WP Carey School of Business (four students), an honors college thesis student in Business Information Systems, and two first year students pursuing independent studies in Computer Science and Engineering. High school students: Directly involved fivehighschool seniors who were exposed to research topics associated with the project as part of their senior projects. Stakeholders in the agricultural products and food supply chain: Outreach involved: Extensive interactions with the International Fresh Produce Association, including a presentation to their Produce Identification Committee and a presentation on NIFA-funded work during their Produce Research Center event at the Global Produce and Floral Show; discussions with the Canadian Produce Marketing Association; discussions and sample transfer with key players in the labeling and packaging space, including Avery Dennison, Bedford, and Sinclair International; visits to Tanimura & Antle's Yuma farming/harvesting operation, Peddlar's Son's produce distribution facility in Phoenix, and the ASU Polytechnic experimental vertical farming operation to plan and execute pilots; presentation to the Arizona Environmental Health Association on secure track and trace in the food industry; introduction of the technology to alarge number of producers and distributors at the Global Produce and Floral Show in Orlando. Scientific community: Connection with the research community via: research update presented to The FoodDistribution Research Society 2021 Annual Conference;delivery of an AAEA Organized Symposium as part of the 2022 Meetings of the Agricultural and Applied Economics Association in Anaheim, July 2022; presentation of two papers at the 9th Annual Conference on Computational Science & Computational Intelligence (CSCI); invited presentation at the 10th International Conference on Advanced Materials and Nanotechnology (New Zealand); presentations at the University of Canterbury (NZ), University of Edinburgh (UK), the Roslin Institute (UK), and University of Oregon; publication of archival journal papers in Advanced in Physics: X, IEEE Access, and Optics Express; paper in preparation for 2024 Agricultural and Applied Economics Association (AAEA) meeting/Agricultural Finance Review. Changes/Problems:There were three significant changes from the proposed plan during the project period: We pivoted from the originally proposed electrochemical technique to a dendrite-forming method involving inks. This changed the experimental plan but yielded excellent results in service of the project goals. Co-PI Gonzalez Velo moved from academia to industry during the second year of the projectso that he was no longer involved in theresearch effort . However, his tasks were handled by Project Director Kozicki and there wasno impact to project timeline or deliverables. We requested and were granted a no-cost extension of the project by six months (the end date wasextended from 12/31/2022 to 6/30/2023). This extension was driven by two factors: (i) The Covid epidemic resulted in issues relating to access to operating farms, which delayed both the fieldstudies and the design of the pilot programs which depended on these. (ii)The Covid epidemic, coupled with the physical move of the computer science research operation from Northern Arizona Universityto Clemson University, delayed the software development portion of the project and slowed the hiring of students. What opportunities for training and professional development has the project provided?This project provided exceptional opportunities for students to gain experience in issues and solutions pertaining to the global agri-food supply chain. Most importantly, these students included many who were not normally associated with agriculture and agribusiness, as well as those who were already studying in these fields. Graduate and undergraduate students in seven disciplines (electrical engineering, materials science and engineering, computer science, management of technology, industrial design, business information systems, and agribusiness), including two female minority students, were directly involved in research activities. Students in honors and capstone projects in engineering, design, agribusiness/business, also were able to become engaged with the project work through thesis work, and we were delighted to get several high school students to perform significant research activities in support of the project. In total, we provided practical training and professional development to more than three dozen students during the project period. How have the results been disseminated to communities of interest?In addition to the thesis of the agri-business PhD student (expected 2024), and thesenior theses and reports produced as a direct result of the experience of the undergraduate students, the main dissemination routes have been direct contact with stakeholders (interviews/discussions, presentations, site visits), several archival journal papers, and presentations at relevant meetings and conferences. Stakeholder interaction included extensive interactions with the International Fresh Produce Association, the Canadian Produce Marketing Association, Avery Dennison, Bedford, Sinclair International, Tanimura & Antle, Peddler's Son, ASU Polytechnic experimental vertical farm, and the Arizona Environmental Health Association. Key presentations included an AAEA Organized Symposium as part of the 2022 Meetings of the Agricultural and Applied Economics Association, a presentation at the International Fresh produce and Flower Show, presentations at the University of Canterbury, New Zealand, the University of Edinburgh, UK, the University of Oregon, twopapers at the 9th Annual Conference on Computational Science & Computational Intelligence, and a paper at the 10th International Conference on Advanced Materials and Nanotechnology in New Zealand. Journal publications includedarchival journal papers in Advancesin Physics:X,Optics Express, IEEE Access, two scientific papers published in arXiv, and a research update in the Journal of Food Distribution Research. What do you plan to do during the next reporting period to accomplish the goals?
Nothing Reported
Impacts
What was accomplished under these goals?
SUMMARY: Working with industry partners and through commercialization efforts, we are now able to produce supply chain compatible Dendritic Identifiers on label substrates at scale. The labels are read using a robust scheme based on an intuitive cell phone app linked to server-based image processing and a label database that links to the product database of choice. The patterns may be authenticated using a novel technique to prevent the fraudulent copying of labels. Initial economic analysis suggests that the DI approach would provide benefits over existing mitigation of supply chain disruptions. 1 Tag creation 1.1 Pivoted to a Dendritic Identifier (DI) formation method involving a patent-pending dispense-compress-separate (DCS) technique that relies on the Saffman-Taylor effect in viscous fluids. Solved issue of dispensing microliter volumes of commonly available ink materials accurately and consistently using positive displacement systems with droplet-substrate contact. Droplet compression by rolling or stamping shown to be suitable for the scalable production of DI patterns of various sizes and densities. Model developed to give dependencies of pattern size and morphology on fluid characteristics and DCS parameters. Optimal DI material and formation parameters determined for scalable manufacturing of high entropy, easily-read DI labels. 1.2 Optimal formulations of agri-food compatible acrylic ink created to give excellent mechanical characteristics and good contrast while being compatible with anti-cloning technique based on micro-reflectors. Protective coatings developed to allow DI labels to survive the use environment throughout the supply chain. Standard industrial label testing employed to establish label robustness (labels shown to be at least as durable as existing food labeling). Ink/coating formulations and optimal DI pattern formation techniques used by an industrial partner to demonstrate manufacturability with large batches of labels (10,000+) produced in a pilot line. 1.3 Developed a patent pending technique for preventing cloning of the tags based on the matching of light scattering from embedded mica micro-reflectors in the pattern to multi-angle reference images. This currently operates off-line but can be used to discourage food fraud through the detection of relabeling and for post-incident determination of label authenticity. 1.4 IMPACT: Through attainment of the objective to create a low-cost, mass-manufacturable, item-level, secure and robust tag, growers, processors, shippers, retailers, and consumers now have a physical identifier that they can use to access highly granular food information. This provides a means to increase trust in the food supply, while enabling hyper-targeted rapid recalls that save lives, reduce food waste, and eliminate economic and brand damage. 2 Pattern reading 2.1 Perfected reading software on a deployable cell phone pattern registration front end and a cloud-based virtual machine/database back end. The ASU approach used license-free ORB image processing, with image/keypoint analysis optimized to give high DI recognition reliability over a wide range of reading conditions. Reading scheme used to determine label readability following lab-based environmental testing and in pilot programs. User-ready version of app was produced for iOS (Apple) phones by ASU spin-out company and tested in pilots. 2.2 Dendrite-specific graph-theoretical approach was perfected (also uses cell phone front end and has processing and database functions on a server). Approach involves image processing prior to extraction of keypoints to improve the accuracy of the app, particularly in the presence of visual noise to minimize error rates. Clemson software team implemented improved versions of deep learning algorithms (based on generative adversarial networks) in backend processing to exploit 3D morphological features on dendritic tags to prevent cloning. Final version has an accelerated authentication algorithm that uses compression techniques for volume dendrite authentication. Key features include the elimination of the need for a training dataset, and generalizability via cross-dataset verification. 2.3 IMPACT: Through attainment of the objective to create a high-reliability user-friendly cell phone-based reading system, we completed what was necessary for a farm-to-fork track and trace system, allowing all agri-food stakeholders to access highly granular safety and nutrition information to improve enterprise efficiency and consumer engagement. Benefits materialize with low investment via use of existing infrastructure for label manufacture and use within the supply chain. 3 Use models and pilot design 3.1 Field studies were performed at Sinclair International's label and labeling equipment operation in Fresno, CA, which included a tour of a large citrus packing facility, the ASU Polytechnic Vertical Farm (VF) in Mesa, AZ, and at Tanimura & Antle's (T&A) winter growing area in Yuma, AZ. Led to the design and implementation of two pilot programs to determine technology efficacy. The VF pilot involved tracking of produce using industry-made DI labels and reading software. The app and database was designed to allow the VF researchers to record and monitor individual plant development under specific growth conditions. The T&A field study of leafy green harvesting and pre-ship processing resulted in a use model for lettuce harvesting in which labeling and data input processes would be performed at box level with no additional operating cost or time disruption to harvest operations. The final T&A pilot involved manual attachment of pre-scanned DI labels to filled boxes at a shipping point at Huron, CA, and the rescanning and recovery of labels at the Peddler's Sons' distribution facility in Phoenix, AZ. All labels survived with full readability following multiple handling operations and a 570 mile trip via refrigerated trailer. 3.2 Developed Monte Carlo model that considers risk/return characteristics (net revenue) for a representative firm that implements a DI based traceability system, including the case of a low-probability but high impact event, such as the near total ban on AZ romaine lettuce in 2018. Model considers comparison of revenue risk/return profile using a DI traceability system vs. a hypothetical revenue insurance policy designed to protect a given level of revenue, and is also being improved to include a comparison with existing but more expensive RFID technology. Considering historical revenue dynamics of romaine lettuce, the hypothesis is that the risk/return profile (mean/variance of revenue) would be improved by the adoption of a DI traceability system that reduces the duration and severity resulting from a major revenue disruption caused by a food safety event, and furthermore is comparable/more cost effective than the hypothetical revenue insurance scheme. Work will continue beyond project termination (for publication in 2024) to include refinement in model calibration, and the revenue-enhancing effect of improved conveyance of food provenance information to consumers and the potential impact on the revenue distribution and risk/return profile of the firm. A standard static Net Present Value model assessing ex-ante financial feasibility of implementing a DI based system will be completed as part of this effort. A further extension will focus on the value of information in the context of innovative food safety technology using a Markov Chain approach. 3.3 IMPACT: Through attainment of the objective to better understand the use environment through field studies and pilots, we were able to establish physical and economic viability, and demonstrate the usability and benefits of the DI approach to stakeholders. This has reduced the barrier to adoption of this technology to mitigate the effects of contamination and counterfeiting in the agri-food supply chain.
Publications
- Type:
Conference Papers and Presentations
Status:
Other
Year Published:
2023
Citation:
Michael N. Kozicki, "Dendrites as Digital Triggers - Harvesting entropy from simple systems on the edge of chaos to protect supply chains," 10th International Conference on Advanced Materials and Nanotechnology, February 7 to 10, Rotorua, New Zealand.
- Type:
Other
Status:
Other
Year Published:
2023
Citation:
Michael N. Kozicki, The Universality of Dendrites, Department of Physics special seminar, University of Canterbury, February 14,2023.
- Type:
Other
Status:
Other
Year Published:
2023
Citation:
Michael N. Kozicki, Dendrites in Nature and Engineering, Department of Physics Seminar Series, University of Oregon, Eugene, OR, April 27, 2023
- Type:
Other
Status:
Other
Year Published:
2023
Citation:
Michael N. Kozicki, Digital Identity in Agri-food Supply Chains, Global Academy for Agriculture and Food Systems Seminar, The Roslin Institute, Edinburgh, May 15, 2023.
- Type:
Journal Articles
Status:
Published
Year Published:
2023
Citation:
Xiwen Chen, Hao Wang, Abolfazl Razi, Michael Kozicki, and Christopher Mann, "DH-GAN: a physics-driven untrained generative adversarial network for holographic imaging," Optics Express 31, no. 6, 10114-10135 (2023).
|
Progress 01/01/22 to 12/31/22
Outputs
Target Audience:The target audience reached during the reporting period was: Graduate and undergraduate students: Directly involved at total of 33 university students (including two female minority students) in seven disciplines - electrical engineering, materials science and engineering, computer science, management of technology, industrial design, business information systems, and agribusiness. In addition to the three PhD, two MS, and three undergraduate students supported under the grant, we also reached an Electrical Engineering capstone team (four members), a Computer Science capstone team (five members), two teams (comprising 10 students in total) in the multi-disciplinary "InnovationSpace" capstone program (organized by the Design School), a senior project team in the Center for Entrepreneurship in the WP Carey School of Business (four students), an honors college thesis student in Business Information Systems, and a first year student pursuing an independent study in Computer Science. High school students: Directly involved four high school seniors who were exposed to research topics associated with the project as part of their senior projects. Stakeholders in the agricultural products and food supply chain: Outreach involved the following: Extensive interactions with the International Fresh Produce Association, including a presentation to their Produce Identification Committee and a presentation on NIFA-funded work during their Produce Research Center event at the Global Produce and Floral Show in Orlando, October 2022; ongoing discussions with the Canadian Produce Marketing Association on the technology; discussions and sample transfer with key players in the labeling and packaging space, including Avery Dennison, Bedford, and Sinclair International; visits to Tanimura & Antle's Yuma farming/harvesting operation and the ASU Polytechnic experimental vertical farming operation to discuss technology and plan pilots; presentation to the Arizona Environmental Health Association on secure track and trace in the food industry; introduction of the technology to alarge number of producers and distributors at the Global Produce and Floral Show in Orlando, October 2022. Scientific community: Connection with the research community via: delivery of an AAEA Organized Symposium as part of the 2022 Meetings of the Agricultural and Applied Economics Association in Anaheim, July 2022; presentation of a paper at the 9th Annual Conference on Computational Science & Computational Intelligence (CSCI), Las Vegas, December 2022; publication of an archival journal paper in IEEE Access; two scientific papers published in arXiv; publication of a research update in the Journal of Food Distribution Research. Changes/Problems:One significant change within this reporting period was the no-cost extension of the project by six months (the end date was extended from 12/31/2022 to 6/30/2023). This extensionwas driven bytwo factors: 1. The Covid epidemic resulted in issues relating to access to operating farms, which delayed both the use environment field studies and the design of the pilot programs which depended on these. 2. The Covid epidemic, coupled with the physical move of the computer science operation from Northern Arizona University to Clemson University, delayed the software developmentportion of the project and slowed the hiring of students. The six month extension was sufficient to get all work back on track. The other notable change was the move of co-PI Gonzalez Velo from academia to industry so that he was not involved in the research effort for most of the reporting period. However, his tasks were handled by Project Director Kozicki and there was no significant impact to project timeline or deliverables. What opportunities for training and professional development has the project provided?This project provided exceptional opportunities for students to gain experience in issues and solutions pertaining to the global agri-food supply chain. Most importantly, these students included many who were not normally associated with agriculture and agribusiness, as well as those who were already studying in these fields. Graduate and undergraduate students in seven disciplines (electrical engineering, materials science and engineering, computer science, management of technology, industrial design, business information systems, and agribusiness), including two female minority students, were directly involved in research activities. Students in honors and capstone projects in engineering, design, agribusiness/business, also were able to become engaged with the project work through thesis work, and we were once again delighted to get four high school students to perform significant research activities in support of the project. In total, we provided practical training and professional development to a total of 37 students during the reporting period. How have the results been disseminated to communities of interest?In addition to the senior theses and reports produced as a direct result of the experience of the undergraduate students, the main dissemination routes have been direct contact with stakeholders (interviews/discussions, presentations), several archival journal papers, and presentations at relevant meetings. Stakeholder interaction included extensive interactions with the International Fresh Produce Association, the Canadian Produce Marketing Association, Avery Dennison, Bedford, Sinclair International, Tanimura & Antle, ASU Polytechnic experimental vertical farming, and the Arizona Environmental Health Association. Key presentations included an AAEA Organized Symposium as part of the 2022 Meetings of the Agricultural and Applied Economics Association, a presentation at the International Fresh produce and Flower Show, and a paper at the 9th Annual Conference on Computational Science & Computational Intelligence. Journal publications included an archival journal paper in IEEE Access, two scientific papers published in arXiv, and a research update in the Journal of Food Distribution Research. What do you plan to do during the next reporting period to accomplish the goals?The plan for the final phase of the project involves three main elements, which align with the original project goals and supporting objectives, as described below. 1 Tag creation Now that we have established ink formulation and choice of label substrate for readable, secure, and robust tags, the key task moving forward is to complete the translational aspects of this project goal. This will involve further IP protection in the form of patent applications and support for the further development of a low cost scalable manufacturing process at our partner company. We fully expect to have mass manufactured tags available for technology commercialization by the end of the project period. 2 Reading system In parallel with the development of mass manufacturing processes for the tags, we will finish and fully test two versions of the reader/authenticator app. The first of these will be a "commercial grade" cell phone-based app that will use available off-the-shelf keypoint detection and analysis software. We already have an operational version of this, but work needs to be performed on the usability/user friendliness and accuracy aspects before any commercial release can be contemplated. We will also complete the light scattering authentication functionality, which is currently carried out in an offline fashion. In addition, we will complete and perfect the second app which employs the graph theory-based custom algorithm to optimize efficiency over the first app, i.e., lower processing costs and decreased response time. In addition, the deep learning approach to identifier authentication will also be perfected and fully demonstrated. 3 Use models The major tasks remaining under the implementation goal are the execution of the vertical farm pilot (P1) and, circumstances allowing, the running of the pilot with Tamimura & Antle (P2). For P1, the principal outcomes will be a demonstration of the use of the Dendritic Identifiers for research data gathering and the efficacy of the technology in a short (emulated) post-harvest supply chain. For P2, determination of the survivability of the identifiers from farm to retailer will be the main outcome. In addition to running and gathering data from the pilots, the economic models will also be completed and analyzed, including the value of information determination.
Impacts
What was accomplished under these goals?
1 Tag creation 1.1 One of the most challenging aspects of working with high viscosity fluids such as our Dendritic Identifier (DI) inks is how to dispense microliter volumes accurately and consistently. We solved this problem by transitioning to positive displacement micropipettes coupled with a droplet-substate contact action to draw the droplet onto the surface via surface tension effects. 1.2 We were successful in creating an optimal formulation of acrylic ink that would give acceptable contrast in DIs formed using our new process while being transparent enough to allow good light reflection from embedded micro-reflectors. This ink formulation was coupled with a high-performance vinyl-based adhesive label substrate and the combination produced DI labels that were easily read. This substrate could also be laser printed so that other patterns could be applied alongside the DIs. 1.3 The label substrate is extremely durable, but to protect the acrylic-based identifier patterns from the environment, various protective materials were tested. The most successful of these was a spray-deposited topcoat of UV-resistant acrylic protector. The finished labels were subjected to a variety of standard label tests, involving heat/cold, moisture, UV exposure, scoring, delamination, and abrasion, and the results indicated that the coated DI labels would survive well in the use environment. We also examined laminated labels, using an adhesive-backed cellulose acetate film as the protective layer, and these appeared to be similarly durable. Neither protective approach negatively impacted the readability of the labels, and the film provided a benefit in the form of an anti-reflection function that eliminated reading errors caused by specular reflections. 1.4 The progress made in the ASU laboratory allowed a rapid transition to an industrial setting. Using the material combination described above, significant quantities of DI labels were produced on an industrial pilot line in a tier-one label manufacturer. Pattern quality and consistency were very good, and the identifiers could easily be recognized using the reader software developed in this project. 1.5 The impact of the work on tag/label creation is that we now envision a scalable scheme for making DI tags that meet all supply chain requirements. This is a major achievement and has driven the overall program forward so that it may achieve the broader societal impact of granular track and trace. 2 Reading system 2.1 We perfected the reading software based on a cell phone pattern registration/field reading front end and a cloud-based virtual machine/database back end. The pattern recognition element in this approach is the commercially available ORB functionality in OpenCV, with keypoint analysis parameters adjusted to give high DI recognition reliability. This reading scheme was used to test the readability of the DI labels produced. As part of the technology translation effort, a commercial version of this app was produced for iOS (Apple) phones by the spin-out company and is currently being tested with potential customers. 2.2 The dendrite-specific graph-theoretical approach to label reading was perfected and implemented. This also uses a cell phone-based front end to capture the initial image of the dendrite and has the processing and database functions on a server. This approach involves image processing prior to extraction of keypoints to improve the accuracy of the app, particularly in the presence of visual noise. This is functioning well and has been tested on ASU- and industry-produced labels with minimal error rates. In addition, the Clemson team has taken a different approach to label authentication by implementing improved versions of deep learning algorithms in the backend processing to exploit 3D morphological features on dendritic tags. The algorithm is based on generative adversarial networks to reconstruct the 3D facets of the dendrites. Additionally, an accelerated authentication algorithm was developed that uses compression techniques for volume dendrite authentication. 2.3 The impact of this work is that we now have a physical reader for DI tags using a cell phone linked to a server that contains a pilot product database. We will use this in the pilots to drive the overall program forward to achieve the broader societal impact of a consumer usable system. 3 Use models 3.1 Significant progress was made in our understanding of the use environment by field studies at the ASU Polytechnic Vertical Farm (VF) and at Tanimura & Antle (T&A) with the VP of Harvest Operations. The VF pilot (P1) involves the tracking of produce using industry-made DI labels. The app and database in this case is designed to allow the VF researchers to record and monitor individual plant development under specific growth conditions as well as provide relevant information to supply chain stakeholders after harvest. This produce is not sold in the open market, but we will emulate a short supply chain by having students and faculty act as growers, distributors, retailers, and customers with label scanning being carried out at each stage. The T&A field study was delayed due to the INSV outbreak in the Salinas Valley; we had to wait until Yuma operations were fully up-and-running in December. We observed how DI labels could be applied in a realistic/real-time lettuce harvest scenario and identified that the existing labeling and data input processes would be easy to adapt to DI implementation at the box level with likely no additional operating cost or time disruption to harvest operations. Adjustments would need to be made to include DIs at the consumer unit level with additional costs to the grower/packer. Implementing the box-level approach with full automation was deemed to be beyond the scope and budget of this project and so we designed a simpler alternative for the pilot (P2) that relied on the manual attachment of pre-formed labels by harvester operators to boxes that were being shipped to a known location (e.g., a warehouse store) so that they could be tracked by project personnel with minimal disruption to the harvesting and distribution operation. The plan was to implement P2 in a California harvesting operation but this was also delayed due to adverse weather conditions in the region. We still fully intend to implement a version of P2 with meaningful data generated on tag survivability and use before the project ends. Extremely useful information relevant to technology development and deployment was also gleaned from visits to Sinclair International's label and labeling equipment operation, which also included a tour of a large packing facility during the processing of citrus crop, and contact with the International Fresh Produce Association and attendance at their Global Produce and Floral Show. 3.2 As part of the associated agri-economics work, we started development of a spreadsheet model that outlines the fixed and variable costs of implementing a Dendritic Identifier for a grower of leafy greens, coupled with a Monte Carlo simulation (using @RISK simulation software) that considers the financial impact from a major disruption due to a food safety scare, and also considers the "speed" of information flow and how it may help mitigate the impact of the scare (e.g., more focused recalls). We will also consider how more information to consumers may impact repeat purchases. We have also started work on an extension of this to determine the value of information in the context of innovative food safety technology using a Markov Chain approach. 3.3 The impact of this work was that we now have a much greater appreciation of the use environment for our innovation and how the technology can be implemented in two very different farming scenarios. We are also making progress on our understanding of the economic impact on stakeholders and society.
Publications
- Type:
Other
Status:
Published
Year Published:
2022
Citation:
Hao Wang, Xiwen Chen, Abolfazl Razi, Michael Kozicki, Rahul Amin, Mark Manfredo, Nano-Resolution Visual Identifiers Enable Secure Monitoring in Next-Generation Cyber-Physical Systems, arXiv: 221.08678 [cs.CR], https://doi.org/10.48550/arXiv.2211.08678 (2022).
- Type:
Other
Status:
Published
Year Published:
2022
Citation:
Xiwen Chen, Hao Wang, Abolfazl Razi, Michael Kozicki, Christopher Mann, DH-GAN: A Physics-driven Untrained Generative Adversarial Network for 3D Microscopic Imaging using Digital Holography, arXiv:2205.12920 [cs.IR], https://doi.org/10.48550/arXiv.2205.12920 (2022).
- Type:
Journal Articles
Status:
Published
Year Published:
2022
Citation:
Huayu Li, Xiwen Chen, Zaoyi Chi, Christopher Mann, Abolfazl Razi, Deep DIH: single-shot digital in-line holography reconstruction by deep learning , IEEE Access, 8, 202648-202659 (2022).
- Type:
Journal Articles
Status:
Published
Year Published:
2022
Citation:
Michael Kozicki, Mark Manfredo, Abolfazl Razi, Yago Gonzales Velo, Research Update: Cell Phone Readable Dendritic Identifiers: Applications for Secure Track and Trace in the Food Supply Chain, Journal of Food Distribution Research, Vol. 53, p9-10 (2022).
- Type:
Conference Papers and Presentations
Status:
Other
Year Published:
2022
Citation:
Michael Kozicki, Trees and Stars - A Novel Approach to Secure Track and Trace in the Agri-Food Supply Chain, AzEHA 2022 Conference, Online, March 9, 2022.
- Type:
Conference Papers and Presentations
Status:
Other
Year Published:
2022
Citation:
Canales, E., Gray, A., Grebitus, G., Kozicki, M., Manfredo, M., and Muth, M. Digital Technology Innovation in Food and Agriculture Food Traceability and Beyond, Organized Symposium (Manfredo, M. Organizer), 2022 Meetings of the Agricultural and Applied Economics Association, Anaheim, CA, August 1, 2022.
- Type:
Conference Papers and Presentations
Status:
Other
Year Published:
2022
Citation:
Michael Kozicki, Mark Manfredo, Abolfazl Razi, Dendritic Identifiers: Connecting Items in the Agri-food Supply Chain to their Digital Presence in the Cloud, Global Produce and Floral Show, International Fresh Produce Association, Orlando, Florida, October 20-22, 2022.
- Type:
Conference Papers and Presentations
Status:
Other
Year Published:
2022
Citation:
Hao Wang, Xiwen Chen, Abolfazl Razi, Rahul Amin, Fast Key Points Detection and Matching for Tree-Structured Images , 9th Annual Conf. on Computational Science & Computational Intelligence (CSCI), Las Vegas, December 2022.
- Type:
Theses/Dissertations
Status:
Published
Year Published:
2022
Citation:
Eleanor Min, "Dendritic Identifiers and Applications in the Food Supply Chain: Understanding the Use Environment of Commercially Grown Leafy Green Vegetables," Honors Thesis, Barrett, The Honors College, Arizona State University, April 11, 2022. https://hdl.handle.net/2286/R.2.N.165223
|
Progress 01/01/21 to 12/31/21
Outputs
Target Audience:The target audience reached during the reporting period is: Graduate and undergraduate students in five disciplines (electrical engineering, computer science, materials engineering, agribusiness, physics). In addition to the students supported under the grant, we also reached three electrical engineering capstone teams (four members each), three teams (comprising 15 students) in the multi-disciplinary "InnovationSpace" capstone program (design, engineering, and business), and three honors college thesis students in agribusiness, engineering, and physics. A high school student in a high acheievement program who performed a theoretical analysis as part of his senior thesis. Stakeholders in the agricultural products and food supply chain, via a research update presented to The Food Distribution Research Society 2021 Annual Conference. The work was also extensively discussed with individual companies and organizations, including producers, distributers, and retailers, and potential manufacturing/commercialization partners. The scientific community, through a review paper in a prestigious scientific journal on the information theory aspects of the work. Changes/Problems:A major component of the proposed work was (and still is) research into optimal material and fornation parameters for the high speed production of well-defined, highly ramified (fractal dimension around 1.7), stable, cm-scale dendritic patterns on standard label materials. Key metrics are dendrite fractal dimension, clarity/contrast (which relates to low error rate readability), and speed of formation, as well as the cost per tag attainable with the chosen process. The original electrochemical process was fairly slow (a few seconds per pattern) and was cumbersome, involving absorbent coatings, electrolytes and electrodes, as well as an electrical stimulation for the growth. During the project work in the first quarter, we realized that we would be unlikely to attain the target specifications while maintaining low production cost and so we switched our focus to a material system that has far greater promise based on the Saffman-Taylor effect in viscous fluids. In this scheme, we compress a few microliters of fluid, typically an acrylic medium mixed with microscale metal flakes, between two surfaces and then separate the surfaces to form a dendritic pattern in the fluid, which is then allowed to harden into a tough but flexible material with light scattering properties that make it very difficult to copy or forge. This is both simple and low in cost and lends itself to roll-to-roll volume manufacturing (note that the materials are used extensively in the paint industry and are very inexpensive and the formation technique is similar to a variety of existing printing processes). This approach changed the scientific approach to the formation of the dendritic identifiers, but did not change any of the goals and proposed outcomes of the project. What opportunities for training and professional development has the project provided?This project provided exceptional opportunities for students to gain experience in issues and solutions pertaining to the global agri-food supply chain. Most importantly, these students included many who were not normally associated with agriculture and agribusiness, as well as those who were already studying in these fields. Graduate and undergraduate students in five disciplines (electrical engineering, computer science, materials engineering, agribusiness, physics) were directly involved in research activities. Students in honors and capstone projects in engineering, physics, design, agribusiness/business, also were able to become engaged with the project work through thesis work, and we were particularly delighted to get a high school student to perform significant research activities in support of the project. In total, we provided practical training and professional development to around 30 students during the reporting period. How have the results been disseminated to communities of interest?In addition to the theses and reports produced as a direct result of the experience of the undergraduate students, the main dissemination routes have been direct contact with stakeholders (interviews/discussions, presentations), and an archival journal paper (Michael N. Kozicki (2021) Information in electrodeposited dendrites, Advances in Physics: X, 6:1, DOI: 10.1080/23746149.2021.1920846) and a conference presentation (Mark Manfredo et al. (2021) Cell Phone Readable Dendritic Identifiers - Applications for Secure Track and Trace in the Food Supply Chain, Research Update, Food Distribution Research Society (FDRS) Conference, Dallas/Fort Worth, Texas, October 17-19, 2021). What do you plan to do during the next reporting period to accomplish the goals?The plan for the next phase of the project involves three main elements, which align with the original project goals and supporting objectives, as described below. 1. Dendritic pattern optimization We have determined the essential materials properties and formation methods that allow us to create secure dendritic patterns at high speed and low cost. The next stage of the experimental work will be to optimize the material composition and processing conditions with a view to scalability and mass manufacturing of the Dendritic Identifiers. This will not only involve ongoing laboratory work that includes material and pattern characterization for a variety of input conditions but also consultation with potential manufacturers. We have already started these conversations and have the strong interest of several ideal partner companies. 2. Reader/app development We are in the process of designing and implementing two different software-based reading approaches. One is the original graph theory-based methodology that is being developed by our partners at Clemson University (the co-PI moved from Nothern Arizona University to Clemson during the first year of the project). This is currently being redesigned to be optimal for the new dendritic pattern formation method and will be completed and tested early in the next reporting period. The other approach utilizes a standard "off-the-shelf" computer vision approach (specifically OpenCV/ORB) to allow us to demonstrate and pilot the technology very quickly. This will probably not be used in the final implementation as it is not as efficient as the custom graph theory but it is simple enough to deploy immediately. 3. Pilot design and implementation As detailed in the original project description, we will pilot the technology under "real world" supply chain conditions. The pilot will be performed in two stages: P1 - This will be limited to vertical farming operations on the ASU Polytechnic Campus and will involve several hundred tags, made in our labs or in partnership with a manufacturer. This is a good initial pilot as the growth environment is highly controlled and the supply chain is short and uncomplicated. The produce (microgreens and strawberries) is also very high value. This pilot will be launched in the summer of 2022. P2 - This will involve a large producer of leafy greens (discussions have started on this) and will require several thousand tags, produced by a manufacturing partner (discussions are also underway). We'll use what we learn in P1 to drive this program, which will be launched in fall 2022.
Impacts
What was accomplished under these goals?
The food supply chain is plagued by the scourges of contamination and counterfeiting; both proliferate in this complex environment and have risen to the level where their effects threaten everyone on the planet as well as inflicting huge economic harm on agri-food stakeholders. Advanced information management, such as blockchain, will help reduce the impact of tainted and fake food but what is missing is an incorruptible physical identifier that provides a secure and highly granular means to link products in the real world to their digital twins in a protected database. We are developing such a secure fingerprint, the Dendritic Identifier (DI), that can be scanned throughout the supply chain using nothing more than a cell phone. The societal impact of this technology could be very great indeed, with an overall improvement in the human condition occurring due to increased trust in the food supply. Individual item identification will lead to hyper-targeted rapid recalls that save lives, reduce food waste, and eliminate economic and brand damage. These benefits materialize with low investment via use of existing infrastructure for label manufacture and use within the supply chain. Reading and authentication of the identifiers by cell phone not only eases adoption and reduces cost, but also brings information all the way to consumers and their social network, allowing and encouraging the socialization of food information and safety. We have achieved much during the reporting period. The impacts and key outcomes associated with the progress toward the original project objectives and the results supporting these are described below. 1. The creation of a low-cost robust and unclonable food-safe tag. We followed the original laboratory research plan in the proposal and developed an electrochemical scheme to create patterns with the desired geometric characteristics and inherent high security. However, we realized that it would be more difficult to transition this approach to a high-speed high-volume manufacturing process than first anticipated due to the large number and conflicting nature of some of the input variables. So, to avoid the potentially long delays associated with the development of the electrochemical concept and the effect this would have on the overall project schedule, we pivoted the work to a much simpler way to form dendrites using a "Saffman-Taylor" (ST) effect in viscous fluids, which could be achieved using a process similar to printing. We had considered this approach in the past but originally ruled it out as we were concerned about the lack of security of the ST patterns. Our new research direction revealed the key material properties (including chemical composition, viscosity, and surface energy) that would yield highly suitable dendritic patterns with the desired contrast, fractal dimension (shape), feature size, and durability. More significantly, we were also able to add a highly secure attribute to these elements by introducing micro-reflectors into the base material that would create authenticating "constellations" - patterns of light that changed with viewing or illumination angle - that could be read by an unaided cell phone. All materials used are common in the paint and printing industries and so are available in bulk at very low cost. The authentication process was so novel and impactful that a patent application was filed. We also demonstrated fabrication of multiple patterns on a variety of label substrates in a printing-style process, validating potential for high-speed mass manufacturing. The substrates included those used for PLU stickers, potentially allowing new and existing technologies to be co-deployed during piloting. The immediate impact of this work is that we now have a scheme for making Dendritic Identifier tags that meet all supply chain requirements. We are able to deploy them in upcoming real world pilots to drive the overall program forward to achieve the broader societal impacts described above. 2. The development of a high reliability pattern reading and recognition system. Halfway through the initial reporting period we elected to pursue two paths toward the development of the cell phone-based Dendritic Identifier reader. The first follows the originally proposed work concerning the graph theory-based approach, which was shown to function well for radial dendrites. This used a newly developed Android OS app with a user interface which captures the patterns and links to a pilot version of web-based back-end image recognition and authentication software in a MATLAB environment. Since this would require more work to function well with the non-radial dendrites produced by the new ST formation method, we also began co-development of a different back-end image recognition system based upon off-the-shelf computer vision software (ORB) which performs feature matching. Initial trials look promising, and the final version will be used in the first pilot program. The immediate impact of this work is that we will shortly have a physical reader for Dendritic Identifier tags using a cell phone linked to a server that contains a pilot product database. We will use this in upcoming real world pilots to drive the overall program forward to achieve the broader societal impacts described above. 3. The design and piloting of stakeholder-driven use models that fit with the economics and physical realities of the food and agricultural supply chain. A major goal of this project is to consider the financial feasibility of adopting the DI technology at various points in the supply chain for fresh produce. In support of this, we sought to thoroughly understand the potential use environment of the technology at the farm, distributor, and retail levels. Such understanding helped to condition a potential test of the DI technology as well as help determine the appropriate cash flows to use in a firm-level net present value (NPV) analysis. This analysis also considers any cost-savings and/or revenue preservation resulting from the use of the DI system in the presence of a food safety scare or recall event. That is, if the information from the DI system can be used to better isolate the cause, this should reduce the financial severity of a recall to firms along the supply chain. While there is obviously considerable heterogeneity among individual farm and food retail environments, providing model use environments will help identify key implementation costs and/or savings as well as a better understanding of how DIs could incrementally fit into current inventory and/or food safety systems. A main outcome of the reporting period is the development of representative use environments for fresh produce. This part of the project involved direct contact with food and agribusiness stakeholders to better understand the use environment, current labeling systems used, and food safety protocols. The conclusion was that we would start our pilot program with vertical farm operations (pilot P1), as the use environment is relatively simple and restricted and the supply chains are short, transitioning to a major producer (pilot P2) using the lessons learned in P1. In addition to understanding the cash flows associated with adoption of DIs, another consideration with respect to overall financial feasibility of the technology is to consider the costs/benefits of DIs relative to competing technologies such as RFID. Towards this end, we conducted a literature search on the adoption of RFID technologies that can be further used to condition a comparative cost analysis. The research team advanced a model that incorporates the findings from the use environment modeling efforts, coupled with a comparative cost analysis of RFID, in developing a simulation model that considers the costs and benefits of the DI technology considering the probability of a food safety scare vis-à-vis the potential cost savings of more hyper-focused recalls.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2021
Citation:
Michael N. Kozicki (2021) Information in electrodeposited dendrites, Advances in Physics: X, 6:1, DOI: 10.1080/23746149.2021.1920846
- Type:
Theses/Dissertations
Status:
Published
Year Published:
2021
Citation:
J. Tensuan (2021) Computational Modeling of the Saffman-Taylor Instability in Lifting Hele-Shaw Cells, BASIS Chandler Thesis, May 2021.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2021
Citation:
Mark Manfredo et al. (2021) Cell Phone Readable Dendritic Identifiers - Applications for Secure Track and Trace in the Food Supply Chain, Research Update, Food Distribution Research Society (FDRS) Conference, Dallas/Fort Worth, Texas, October 17-19, 2021
- Type:
Theses/Dissertations
Status:
Published
Year Published:
2021
Citation:
S. Robertson (2021) Robotics Based Durability Testing of Dendritic
Identifiers for a Secure Produce Supply Chain, Barrett Honors College Thesis, Arizona State University, May 2021.
- Type:
Theses/Dissertations
Status:
Published
Year Published:
2021
Citation:
C. Bradshaw, et al. (2021) Dendritic Identifier Durability Testing, Electrical Engineering Senior Design Final Report, Arizona State University, May 2021.
|
|