Lygodium microphyllum, commonly known as Old World Climbing Fern (OWCF), is now a prevalent invasive species in the state of Florida. The plant is native to the Eastern Hemisphere, including regions of tropical and south Africa, tropical Asia, Australia, and Oceania. First discovered in the wild in Florida around 1960, it became well established by 1978. Since it can reproduce through wind-dispersed spores, the plant can rapidly spread from existing populations. Land area covered by OWCF increased nearly fivefold in the state of Florida from 27,000 acres in 1993 to over 120,000 acres in 2005, and has continued to increase. The UF/IFAS declares it �������one of Florida��������s most detrimental and difficult to manage invasive plants�������. The impact of OWCF is multi-fold: it can grow up and over existing vegetation, forming a canopy that deprives trees and understory vegetation of light, and forming a dense mat of vegetation the ground. The impact of this growth can be lethal to the understory vegetation and even mature trees, eliminating these native plants. It has the ability to grow in a variety of plant communities, including bald cypress stands, pine flatwoods, wet prairies, saw grass marshes, mangrove communities and Everglades tree islands. For a location with difficult access, the cost of chemical treatment exceeded $3,500 per hectare in the year 2000. OWCF also imposes a fire hazard, as the vegetation can serve as a fire ladder to propagate flames upward into the tree canopy. It can also spread fire horizontally as burning vegetation becomes detached. While mechanical treatment is often infeasible, biological and chemical methods have been employed to a more significant degree to combat the growth of the fern. Chemical herbicides have proven efficacious if treatments are repeated over a multi-year period. For example, bi-annual treatments over a two-year period resulted in over 96% reduction in fern coverage. For large concentrations of OWCF, helicopters are often used for herbicide treatment, while isolated plants are usually spot-treated with handheld or backpack sprayers. The use of small Unmanned Aerial Vehicles (sUAVs) for mapping is now commonplace in a variety of industries. Perhaps the most common form of sUAV-based mapping involves autonomous GPS navigation and image capturing through unobstructed areas, predefined by the user. Such functionality can be readily achieved with commercial off-the-shelf (COTS) hardware and software, e.g., DroneDeploy. Though the literature is replete with examples of remote-controlled sUAV obstacle avoidance, including both high-speed swarm travel through forests, sUAVs with the ability to avoid obstacles during autonomous mapping are rare. The Emescent Hovermap LiDAR unit has been used in conjunction with a COTS sUAV for simultaneous localization and mapping (SLAM) in GPS-denied areas. Scion, a New Zealand company, has demonstrated the coupling of a commercial sUAV with the Hovermap for SLAM under a tree canopy. A major drawback is the cost of this system, reportedly more than $60,000. The proposed research seeks to develop and demonstrate a low-cost sUAS capable of autonomous monitoring and management of OWCF under the cypress canopies of central Florida. Year 1 will focus on the development of an autonomous sUAV with a high resolution RGB camera for SLAM with obstacle avoidance, and imaging of plants below forest canopies. Its payload capabilities will be sufficient to transport 1.5 liters of chemical herbicide, with a target flight duration of 30 minutes or greater. A homing feature will return it to the operator when battery levels fall to within 20% of full capacity. The deep neural network (DNN) plant classification algorithms developed by our research team will be trained and implemented to enable real-time identification and monitoring of OWCF in these challenging habitats. Year 2 work will focus on the development of autonomous herbicide application and field validation studies.

Aquatic invasive plants (nonindigenous aquatic nuisance species) may have significant negative consequences upon the water systems that they invade. Hydrilla verticillata, in particular, has invaded a significant portion of the U.S. from Florida to California and as far north as Maine and Idaho. In order to properly assess and manage invasive aquatic plants, managers must determine both the scope of infestation as well as the specific species that present. In recent years, managers have been increasingly using hydroacoustic sensors to quantify submersed plant distributions and densities. This has provided increased efficiency over traditional techniques like point intercept surveys and aerial imagery. However, hydroacoustic sensors and current data processing only provide a representation of �������biovolume������� and do not provide speciation. Therefore, managers must still conduct some level of point intercept or diver effort in order to determine the specific species that are present. These activities are labor intensive and represent large efficiency gains if these activities can be replaced by advanced data processing. Automated plant identification methods, using high resolution true-color and multispectral imagery, have proven effective for land-based plants but have limited application to submersed aquatic vegetation. Despite the shortcomings of traditional techniques, our recent studies have shown that hydroacoustic imagery can be implemented for submersed aquatic species classification, with the use of deep learning, an advanced machine learning technique. Hydroacoustic data was collected via a small fleet of fully autonomous boats with both subsurface hydroacoustic imaging and herbicide deployment (for vegetation control) capabilities. The proposed research will integrate deep neural networks (DNNs) to develop advanced data processing techniques that classify submersed plant species from hydroacoustic data. While research to date has focused on a limited number of plant varieties (Hydrilla, Cabomba, Coontail, or �������other�������) and geographic locations, DNN training will be expanded to a much wider range of invasive species and locations. To minimize misclassification related to plant maturity, hydroacoustic imagery will also be collected throughout the growing season. True-color (RGB) images both above and below the water surface will be included (sensor fusion) to validate DNN classification.

Aquatic invasive plants (nonindigenous aquatic nuisance species) may have significant negative consequences upon the water systems that they invade. Hydrilla verticillata, in particular, has invaded a significant portion of the U.S. from Florida to California and as far north as Maine and Idaho. In order to properly assess and manage invasive aquatic plants, managers must determine both the scope of infestation as well as the specific species that present. In recent years, managers have been increasingly using hydroacoustic sensors to quantify submersed plant distributions and densities. This has provided increased efficiency over traditional techniques like point intercept surveys and aerial imagery. However, hydroacoustic sensors and current data processing only provide a representation of ����������������biovolume��������������� and do not provide speciation. Therefore, managers must still conduct some level of point intercept or diver effort in order to determine the specific species that are present. These activities are labor intensive and represent large efficiency gains if these activities can be replaced by advanced data processing. Automated plant identification methods, using high resolution true-color and multispectral imagery, have proven effective for land-based plants but have limited application to submersed aquatic vegetation. Despite the shortcomings of traditional techniques, our recent studies have shown that hydroacoustic imagery can be implemented for submersed aquatic species classification, with the use of deep learning, an advanced machine learning technique. Hydroacoustic data was collected via a small fleet of fully autonomous boats with both subsurface hydroacoustic imaging and herbicide deployment (for vegetation control) capabilities. The proposed research will integrate deep neural networks (DNNs) to develop advanced data processing techniques that classify submersed plant species from hydroacoustic data. While research to date has focused on a limited number of plant varieties (Hydrilla, Cabomba, Coontail, or ����������������other���������������) and geographic locations, DNN training will be expanded to a much wider range of invasive species and locations. To minimize misclassification related to plant maturity, hydroacoustic imagery will also be collected throughout the growing season. True-color (RGB) images both above and below the water surface will be included (sensor fusion) to validate DNN classification.

Providing reconnaissance and situational awareness in large and complex subterranean facilities will require multiple distributed sensing platforms (e.g. small multi-rotor UAVs) that can distribute themselves and navigate autonomously throughout the space. Implementing such a system poses several design challenges. How can multiple vehicles be inserted into the area with minimal human intervention? How can the swarm intelligently manage power levels across vehicles to provide both rapid initial mapping as well as continuous sensor coverage and endurance? These questions highlight UAS mobility, system power management, and decentralized exploration as key challenges in the development of a rapid subterranean mapping and monitoring system. One potential solution is a man-portable ����������������mothership���������������: an autonomous ground robot that can serve to launch, recover, and re-energize a swarm of small aerial vehicles. The mothership can be designed for rapid deployment through a small doorway or window. If needed it can then traverse to an initial launch point for the UAVs, and can continue to move forward to new locations as needed. When the power level of an individual UAV becomes low, it can return to the mothership for automatic battery recharge or replacement. Other vehicles in the swarm can be reallocated to continue exploration forward or maintain sensing and communication coverage throughout an area to be monitored. The ground robot itself can also provide additional heterogeneous capabilities to the system, including carrying larger sensor packages that exceed the payload capabilities of the UAVs, and perhaps featuring manipulator arms or sample collection equipment to investigate objects of interest. The proposed project will investigate this mobile UAS mothership concept through a combination of trade-off studies, analysis, conceptual design, and prototyping and testing of key subsystem technologies.

Providing reconnaissance and situational awareness in large and complex subterranean facilities will require multiple distributed sensing platforms (e.g. small multi-rotor UAVs) that can distribute themselves and navigate autonomously throughout the space. Implementing such a system poses several design challenges. How can multiple vehicles be inserted into the area with minimal human intervention? How can the swarm intelligently manage power levels across vehicles to provide both rapid initial mapping as well as continuous sensor coverage and endurance? These questions highlight UAS mobility, system power management, and decentralized exploration as key challenges in the development of a rapid subterranean mapping and monitoring system. One potential solution is a man-portable ����������������mothership���������������: an autonomous ground robot that can serve to launch, recover, and re-energize a swarm of small aerial vehicles. The mothership can be designed for rapid deployment through a small doorway or window. If needed it can then traverse to an initial launch point for the UAVs, and can continue to move forward to new locations as needed. When the power level of an individual UAV becomes low, it can return to the mothership for automatic battery recharge or replacement. Other vehicles in the swarm can be reallocated to continue exploration forward or maintain sensing and communication coverage throughout an area to be monitored. The ground robot itself can also provide additional heterogeneous capabilities to the system, including carrying larger sensor packages that exceed the payload capabilities of the UAVs, and perhaps featuring manipulator arms or sample collection equipment to investigate objects of interest. The proposed project will investigate this mobile UAS mothership concept through a combination of trade-off studies, analysis, conceptual design, and prototyping and testing of key subsystem technologies.

Three-dimensional in vitro cell cultures are finding increased application in the study of solid tissues. Both "simple spheriod" cultures derived from single cell types and "organoid" cultures derived from multiple cell types can be readily established using 96-well plates molded from ultra-low attachment substrates. Because these 3D cultures more closely resemble in vivo tissues than their 2D counterparts, they may provide more accurate modeling of in vivo tissues and prediction of patient outcomes. A limiting factor is the histologic analysis of 3D cultures using existing tools and techniques: the manual process is time-consuming and inefficient and cannot compete with the throughput of robotic systems used in the screening in microwell plate formats. This research seeks to overcome these limitations through the development of Smart Material Carrier Basket Arrays (smCBAs) that will enable simultaneous and direct transfer of the spheroids/organoids contained in a 96-well plate into a histology cassette for routine processing and paraffin embedding of the 8 x 12 array as a single specimen. Our proposed smCBAs will be fabricated from laser-cut sheet nitinol, which will be thermally activated to facilitate this transfer.

American automobiles consume 365.7 million gallons of gasoline per day, and international demand for fuel is increasing at a staggering rate. Although revolutionary technologies are on the horizon, improvements to existing engines have the potential to improve efficiency and reduce emissions now, while providing benefits far into the future. Introducing even modest improvements (5-10%) in the efficiency of internal combustion engines has the potential to save millions of gallons of fuel per day with reduced emissions. One technology with the potential to significantly improve the efficiency of internal combustion engines while simultaneously reducing emissions is variable-geometry spray (VGS) fuel injection. This patent-pending technology enables independent control of fuel flow rate and spray geometry as it enters the combustion chamber. With very limited R&D resources, the investigators have developed three generations of VGS fuel injector prototypes, and preliminary results demonstrate device functionality with low-pressure non-combustible fluids. This VGS injection technology contains novel features which promise improvements over existing injection methods: dual computer-controlled actuators which regulate spray geometry and fuel flow rate independently and continuously throughout the injection process. Due to the uniqueness and commercial potential of this technology, patent applications have been filed through N.C. State’s Office of Technology Transfer (International Application #: PCT/US2009/030707, US Application #: US 2011/0005499 A1). This proposal seeks Chancellor’s Innovation Funding to support the fabrication and testing of VGS fuel injector prototypes in an internal combustion engine to validate the device's performance and enhance opportunities for licensing the technology.

The primary objective of this Phase II research will be to design, fabricate, and surgically evaluate a robotic catheter prototype for atrial ablation procedures. Phase I research successfully demonstrated the feasibility of this concept through the development of a working catheter prototype with two highly maneuverable shape memory alloy (SMA) actuated bending segments. Phase II research will build upon the design achievements of Phase I to produce a catheter with full ablation capabilities, enhanced maneuverability, computercontrolled ablation modes, and contact stability during ablative energy application. This technology has the potential to transform such procedures by providing unprecedented maneuverability, visualization, and access to open spaces within the heart.

Advanced fuel injection strategies are central to efforts to improve the performance and emissions characteristics of internal combustion engines. Existing fuel injectors severely restrict the strategies that can be explored, however, because of their fixed spray geometries. We propose to develop and experimentally study a novel fuel injector that can continuously vary its spray angle throughout the injection process, and do so independently of fuel flow rate. Such an injector could increase engine efficiency and reduce pollutants by optimizing air-fuel mixing and the distribution of atomized fuel within the cylinder. Preliminary data (from experimental testing at atmospheric conditions and corresponding numerical simulations) supports the feasibility of our proposed Variable Geometry Spray (VGS) technology, but further testing at realistic engine conditions is needed. Using this VGS prototype, we propose to investigate the transient effects of spray geometry variation on fuel atomization, air-fuel mixing, ignition, and pollutant formation processes at realistic injection pressures and in-cylinder ambient conditions. Our research will experimentally demonstrate computer-controlled, VGS fuel injection and will quantify its impact on the production of NOx and SOx, as well as overall combustion efficiency. We will directly compare the results from our prototype injector to commercially available gasoline direct injectors. Various imaging and measurement techniques will be used to study the structure, droplet size, and mixing characteristics of variable geometry sprays. Chemiluminesence and [other techniques?] will be used to examine pollutant formation. If this research is successful, we expect it will lead to significant near-term improvements in existing IC engines and will facilitate the implementation of new combustion modes such as HCCI and flexible fuel capability.

The primary objective of this Phase I SBIR proposal is to develop and demonstrate shape memory alloy (SMA) actuated catheter technology that will result in the development of robotic catheters for minimally invasive surgery and catheterization in Phase II. This technology has the potential to transform such procedures by providing unprecedented maneuverability, visualization, and access to open spaces within the body. Benefiting from the precision and repeatability of computer-based control, these catheters have the potential to impact a variety of medical fields, including cardiology, cardiac surgery, pediatric surgery and urology.