Recipient Organization
Geo-Spider, Inc
4406 NW 77 Terrace
Gainesville,FL 32606
Performing Department
(N/A)
Non Technical Summary
Numerous challenges face the Florida and U.S. citrus industries, including global market pressure, urban and environmental pressures, labor issues, and emerging diseases and pests, thus underscoring the need for new production approaches. Over the last two decades, over 200,000 acres of Florida citrus production has been lost due to the citrus canker eradication program and citrus greening (HLB). HLB is now considered endemic in Florida and hundreds of thousands of acres are in steep decline, with minimal hope of finding remediation solutions. Long term hope rests on finding HLB resistant varieties and developing advanced management strategies that can control the psyllid population, repel them from groves, and manage trees in an optimal economic framework. One approach being considered is Advanced Citrus Production Systems (ACPS) that use high density semi-dwarfed trees, and open hydroponics with optimized nutrient and water availability, which accelerates plant growth. When combined with HLB tolerant rootstocks, the concept seeks to increase yield production per acre, while simultaneously shortening the time to return on investment, which means that grove life can be shortened by disease pressure and still remain economically viable. Another emerging, commercially admitted production approach is Citrus Under Protective Screen (CUPS); CUPS uses high-density semi-dwarfed trees or potted trees to limit tree size due to canopy height limitations. Citrus under enclosed structures can exclude the Asian citrus psyllid and eliminate the negative effects of citrus greening to high value varieties like the grapefruit. However, both of these new grove architectures will require new compatible equipment systems to manage production and harvesting, but they may create alternate crop disease and pest pressures that require precision spray control to manage.GeoSpider, Inc has been developing both mass and selective harvesting technologies for high density ACPS. However, we recognize that there is a signficant overlap between ACPS and CUPS, not only in harvesting, but also in sprayer related technologies. The citrus industry is facing a challenge with respect to a large number of disease and pests that require the use of expensive chemicals, some of which require monthly application. With growing environmental pressures and concerns over agricultural sustainability, it will be crucial to develop more efficient autonomous technologies that can scout and map disease and pest pressures using artificial intelligence, which can build prescription maps along with real-time canopy mapping to optimize sprayer efficacy and minimize overspray and drift, which will result in environmental and economic benefits for producers.In this Phase I project, we will develop a scaled concept prototype to demonstrate autonomus navigation, canopy mapping, under screen localization, and precision spraying with the following specific objectives: 1) We will adapt an earlier greenhouse robot sprayer rig for demonstration of autonomous navigation under canopy and spraying in CUPS, while 2) a manned JD Gator and a pull behind sprayer will be equipped with the sensory systems for scouting, navigation, canopy mapping, and precision spraying using technologies developed during prior research. In addition, 3) advanced concept design and modeling will be done for the full-scale Phase II robot scout and sensory suite. 4) Field tests will be conducted using the scaled Phase I prototype at the Dundee CUPS facilities. In the planned Phase II project, we will integrate the technology components from prior research and this Phase I project to build a commercial Phase II Robotic Scout and Precision Sprayer for high-density citrus in CUPS. The Phase II prototype can then be tested at the UF ACPS grove in Citra, FL and at the Dundee CUPS facilities. This technology can then be used in high-density citrus production between the rows in CUPS or ACPS, or it can be adapted to the ACPS Over-the-ToP (OTP) platform which straddles the tree in ACPS.
Animal Health Component
25%
Research Effort Categories
Basic
25%
Applied
25%
Developmental
50%
Goals / Objectives
Since the advent of HLB in Florida, the profitability and viability of citrus production has been gradually diminishing due to declining crop yield and quality, and significant increase in production costs due to Asian Citrus Psyllid (ACP, Diaphorina citri) control and Foliar Nutritional Application (FNA) that attempt to compensate for loss of nutritional uptake through damaged root systems. The intensive use of canopy applied psyllid control and foliar nutritional programs demonstrate the need for highly accurate spray delivery systems that only spray tree canopy, and vary application according to within grove variability. These same conditions apply to disease control measures for postbloom fruit drop (PFD) and citrus black spot (CBS). A recent economic analysis by Singerman showed that the annual cultural costs for SW Florida juice orange was $1910/ac/yr, with the cost for spray applications at $630/ac/yr, or 33% of the production costs.Long term hope rests on finding HLB resistant varieties and developing management strategies that can control the psyllid population and manage trees in an optimal economic framework. One management approach being considered is Advanced Citrus Production Systems (ACPS) that use high-density (HD) semi-dwarfed trees and open hydroponics with optimized nutrient and water availability, which accelerates plant growth. Another approach being considered is Citrus Under Protective Screen (CUPS), CUPS uses HD semi-dwarfed trees or potted trees to limit tree size due to canopy height limitations. Citrus production under enclosed structures can exclude the ACP and eliminate the negative effects of HLB caused byCandidatusLiberibacter asiaticus to the grapefruit (Citrus paradisi) fresh fruit industry. Since the discovery of the ACP, the importance of developing advanced production systems has only increased. This is evidenced by the growing number of aces being planted in CUPS. At the end of 2019, Ferrarezi reported that there were in excess of 400 acres of CUPS in Florida, while recent conversations with Dundee Growers report that they alone plan to establish over 1000 acres in CUPS at approximately $40K/acre establishment cost for structure, trees, irrigation, and other infrastructure. Thus, demonstrating the confidence and commitment that some growers have towards CUPS.Both of these production approaches, will require new equipment systems to manage production and harvesting. Autonomous operations and equipment are well suited for both ACPS and CUPS, which both require closely spaced smallish plants. Such trees have known efficiencies in their cultural and harvesting management, and constitute ideal orchard systems to accommodate autonomous scouting, pruning, mowing, spraying, and harvesting. Likewise, many of the same sensor, control and automation/robotic features will be relevant for both applications with the primary difference being the mobile platform design used for CUPS vs. ACPS. CUPS will likely be applied to high value crops (e.g., grapefruit) that can offset the high startup costs. At a smaller production scale with high value fruit, trees can be managed in an optimal fashion for improving spraying and harvesting effectiveness due to reduced tree height and shallow canopy depths grown in a fruiting wall. In addition, the screen provides a natural solar light diffuser for illumination optimization that will assist in optimizing disease and pest scouting as well as robotic harvesting. The shallow canopy depth will improve spray penetration and minimize harvesting obstacles.In this Phase I proposal, we will focus on the building a proof of concept prototype CUPS scout and precision sprayer at an approximate ¼ scale. The eventual Phase II system will consist of three primary sub-systems. First, the Autonomous Scout Powerplant which will be powered by a 50 HP diesel engine with both hydraulic and electric power distribution via a generator and hydraulic pumps. The scout will be capable of running independent of implements during scouting missions being equipped with a suite of situation awareness sensor technologies, which can monitor spatial position within CUPS, map tree canopy profile and volume, guide the vehicle with obstacle avoidance, conduct disease and pest surveillance to assist precision sprayer and map fruit positions to assist selective harvesting. The second primary sub-system is the trailer mounted precision sprayer, which will be powered and controlled by the scout powerplant. It will be equipped with a water supply tank, chemical storage canisters, chemical injectors for custom blending and individual sprayer solenoids for precise application coverage. The initial sprayer, which we will evaluate, will use air-assisted low-volume atomizing technology to penetrate canopy with precision nozzle control. The sprayer will be controlled based on real-time scouting information and prescription maps according to disease/pest pressure provided by the RSPS scout or human experts, and other horticultural factors that may impact spray treatment efficacy. The third subsystem will be the subject of future research for development of the CUPS-based selective harvesting platform, which like the precision sprayer will be powered and controlled by the scout. These three technologies will also be directly applicable for the ACPS sprayer and harvester.The specific objectives of this proposal are as follows: 1) Evaluate current designs, 2) Design and modify RSPS, 3) Integrate sensors and software in sprayer rig, 4) Modify MSPS JD Gator with sensor suit and software, 5) Adapt/retrofit precision sprayer rig on MSPS, 6) Integrate MSPS sensors and software with precision sprayer, 7) Conduct proof of concept trials, 8) Refine Commercialization Plan and Report.
Project Methods
We will seek to build off prior research by Burks and team demonstrating to potential grower users and investors two facets of the systems performance. However, since this is a complicated system and Phase I funds and time allocation are limited. We plan to leverage prior work and implement proof of concept using two separate and parallel physical systems. First, we will adapt our greenhouse robot shown by taking the DC motors and controllers and mounting them in a slightly bigger framework that is more appropriate for CUPS domain. These motors have sufficient torque to power a scaled version of RSPS that will be able to demonstrate indoor autonomous localization and navigation coupled with a preliminary canopy mapper. We will hence forward refer to this as Robotic Scout and Precision Sprayer One (RSPS). Then we will adapt a manned JD Gator with the concept scout sensor suite, DGPS, and a custom Mapping and Sprayer Controller, which will then be coupled with a scaled version of the concept precision sprayer. Henceforth we will refer to this set-up as Manned Scout and Precision Sprayer One (MSPS). By combining real-time sensing (canopy mapping) with mock-up prescription maps, MSPS will be able to demonstrate precision canopy profile spraying with custom chemical injection via individual nozzle control.The two related parallel prototypes will be developed that will independently demonstrate autonomous operation in the CUPS facility, and separately demonstrate the capabilities of scouting and precision spraying both in CUPS and later in ACPS. The learning outcomes of these trials will be used to enhance design stage efforts in development of the Phase II scout and sprayer. The two will be coupled together in Phase II in a pre-commercial prototype RSPS to demonstrate full system potential. Since it is not practical to implement the full autonomous robotic scout and precision sprayer in a Phase I prototype we have decided to use systems that can be more readily adapted to demonstrate the concept of indoor autonomous guidance and canopy scouting in CUPS, and secondly the concepts of precision spraying in ACPS. This way we can perk the attention of both markets. In RSPS, our scaled prototype will demonstrate autonomous capabilities, while the MSPS1 will demonstrate precision spraying. We will take one of our manned Gators and retrofit it with a sensor suite consisting of DGPS for guidance and mapping, multiple machine vision cameras for canopy monitoring, and LiDAR for canopy mapping. The camera systems will be used for monitoring leaf density and canopy volume for the precision sprayer. Later in Phase II, it can be used for more sophisticated disease or pest scouting when coupled with AI-based detection algorithms and appropriate vision technology.We will suitably modify the robot sprayer reported previously to make it more appropriate for trials in a CUPS facility. This robot will be doubled in scale, which will end up being about ¼ scale of the eventual RSPS. RSPS will use the same high torque-low speed DC motors and controllers, but in a new chassis with improved suspension and power management. The eventual Phase II prototype will have articulated steering at the hitch point along with differential front wheel drive to improve mobility, while RSPS will still use a traditional skid steering approach. An optional 4-wheel drive will be possible in Phase II. We will upgrade the embedded motion controller as well as the vision controller to an NVIDIA Jetson Xavier with AI capability. The RSPS will be equipped with on-board 10-gallon sprayer.We will then purchase a medium-sized 50-gallon sprayer with chemical injectors and low volume atomizers. AirTec Sprayers, Inc. has agreed to participate with us on this project and will likely provide valuable pricing discounts and technical assistance. So, choosing to customize an AirTec will provide valuable visibility for our systems and could become a marketing outlet or a potential manufacturing partner. In this case, we will need to provide a secondary power supply for the sprayer independent of the Gator, which will serve as the manned scout and towing vehicle for the sprayer, but it doesn't have sufficient power for the sprayer as well. The Phase II RSPS will provide all necessary power. The sprayer will be equipped with individual nozzle solenoid valves and variable rate injection and variable air assist, which will allow us to precisely apply chemicals to the profile of the tree matched with the canopy leaf density to optimize spray delivery efficacy. At this point we will integrate the sensory system through one or more AI-enabled Nvidia Jetson Xaviers along with a lower-level embedded processor with relay outputs that will run the variable rate sprayer with precision nozzle control. All systems components will be integrated into a ROS operating system which will later be compatible with RSPS in Phase II. The multiple Xavier's will be responsible for housing the MSPS scouting algorithms and collecting the various sensor data that will be used for geo-spatial canopy mapping, canopy volume, profile, and leaf density, reading prescription maps and sending sprayer control measures to the solenoid controller. In this Phase I, we will simply focus on real-time canopy profiling for precision spraying.Once fabrication, retrofit, and integration are completed on both the RSPS and MSPS, the assembled prototypes will be tested. RSPS will first be tested in the laboratory on a mockup test track, and once basic navigation is operational, it will be moved to a UF greenhouse, where the RTLS will be setup and operational. At this point, potted citrus trees will be set up in a CUPS type course so that localization, sensor-based navigation, and basic on-off sprayer control can be tested. This will include positioning accuracy and spray delivery accuracy tests, where performance data will be collected and analyzed. In parallel with these trials, we will be preparing for trials on MSPS, where a local farm will be used for preliminary GPS based mapping of citrus tree canopy and precision spray application based on canopy profile. Once these preliminary trials are complete, we will move the equipment to the UF PSREU campus for further trials in an ACPS grove. In this case, MSPS will be used to map the tree canopy, collect canopy profile, volume and leaf density measures, and run the precision sprayer based on measured data.Performance evaluation will seek to demonstrate whether this novel robotic scout and precision sprayer can successfully navigate in high density citrus in CUPS and provide precision sprayer control based on tree canopy reconstruction in a viable economic framework. At this point, we will invite growers to observe RSPS and MSPS performance, and if practical will take the two prototypes to actual CUPS and ACPS groves for demonstrations. Once all trials are finished, a product evaluation will be conducted through feedback from growers and industry partners to access the performance of the Phase I and revise the commercialization plan. Then the final report will be written.