In March 2016, AlphaGo, an AI program by Google DeepMind, defeated the Go world champion Lee Sedol 4-1 in a five-game match, shocking the world. After March 2017 when AlphaGo again defeated the world champion, AlphaGo was improved to a newer version called AlphaZero, a stronger AI program that self-trained, with no prior knowledge, after being told only the rules of the game. From then, the strength of AI kept climbing at an astonishing rate.
Gian-Carlo Pascutto, a computer programmer who works at the Mozilla Corporation, had a track record of building competitive game engines, first in chess, then in Go. After following the latest research, he combined the Monte Carlo Tree Search and a neural network into building the world’s most successful open-source Go engines – first Leela, then LeelaZero – which mirrored the advances made by DeepMind.
Based on the open-source engines, we plan to take an alternative path of utilizing LeelaZero: finding the optimal results/playouts on different board sizes from 3x3 up to 9x9. Because of symmetry, there is a difference between an even and odd n x n board size. Therefore, we treat them separately on the following outline of the project:
- Modify the code of LeelaZero to allow all odd dimensions and obtain the results of optimal play for odd n up to 9.
- Modify the code of LeelaZero to allow all even dimensions and obtain the results of optimal plays for even n up to 8.

The new TCU and UNTHSC School of Medicine is taking a progressive approach to curriculum for their students. The standard for medical clerkships, is for a medical student to focus on a practice, then move on to the next practice. This leaves a gap of time between learning and implementing a medical practice in the real world. The Longitudinal Integrated Clerkship (LIC) will engage students in a variety of medical practices in 2 week cycles, so students will constantly be maintaining their grasp on import skills and practices. It is our job to provide the scheduling application that will best match each student and doctor, at the best times.

Sheepdog Defense Group is a local Fort Worth self-defense company that is fully licensed by the State of Texas Private Securities Bureau to provide self-defense and weapons training to help other protect their communities. Their main goal is to provide training to church groups and private schools to help them from becoming targets for acts of violence. Sheepdog Defense Group also offers this self-defense and weapons training to the public so that they can protects them selves and their families. Sheepdog Defense Group is looking for a new website that will allow customers to sign-up for classes using an interactive calendar and access an online store to purchase Sheepdog merchandise. The site will allow the Sheepdog Guards to access all of the important information needed to protect their community as well as access their own information. The site will also allow the owner to manage a wide range of services regarding the business which he is currently doing all by hand.

As part of one of the engineering capstone projects, a calibration testing system was improved with the aid of computer vision. Computer vision was integrated into this project as a solution to a rotating pedestal calibration test that was previously performed by the naked eye. The main goal of this system was to detect and track a red 635 nm wavelength laser spot with offsets as small as 0.025 inches on a 10 x 10 inch grid accurately and precisely. Designing this system involved three major criteria: camera selection, data processing hardware, and algorithm performance.
The first criteria studied in the design process was the camera. The system required a camera that was compact in size, covered the entirety of the grid at less than 11 inches, and captured high quality images. Furthermore, two main data processing hardwares were explored: Raspberry Pi and a standard test laptop. The processing hardware criteria considered were speed, portability, and maintenance. Finally, RGB and houghcircles were the two algorithms used to detect the red laser dot. Testing was conducted to compare the algorithms based on their ability to detect the laser spot, precision in tracking, and repeatability. These design considerations guided the down selects for the final components used in this system.

The LabVIEW team for the Applied Avionics Inc. project focuses on fully integrating the programming of all electrical components with LabVIEW. The major requirements for this project include utilizing LabVIEW to display and capture data feedback, completely automate the testing process, and to read and send data directly to AAI’s database. By creating an actuation and extraction feedback machine that is fully LabVIEW controlled, a variety of switch body types were able to be accommodated and tested. The machine has been shown to decrease variability of results and improve the efficiency of AAI’s current process in all aspects required.

Cloud based services such as IBM Cloud and Amazon Web Services provides a new platform for data collection, storage and processing through the internet that enables environment monitoring via wireless sensor networks. In this project, we would like to develop a cloud-based low power monitoring and notification platform using AWS. Most existing notification platforms are provided as an expensive, closed system that do not allow flexibility and is often difficult to troubleshoot. These systems require special hardware (such as unique walkie-talkies) and upgrades are pushed back due to costs.
Our system will utilize AWS Lambda functions, a cloud database, and IOT buttons so that medical staff can receive and store real time patient vitals and notifications with a data forwarding device such as a smart phone, tablet, or computer. AWS solutions are low-cost and flexible, allowing the care centers to customize the functionality to their specific needs. These buttons do not require wired power supply and have a long-lasting battery.

We are presenting a method referred to as Hydrogen Production by HyPIR Electrolysis. The method increases the rate of hydrogen production from a 1 molar potassium hydroxide and water solution under 6 volts when an infrared laser is irradiated with an optimum wavelength of light through a cell and concentrated on exposed copper electrodes. The irradiating light facilitates the dissociation of water by stretching the hydrogen oxygen bonds and increasing the rate of hydrogen production. Production of hydrogen due to the class 4 laser is altered by the specifications of laser energy, pulses per second, and spot size.

In this experiment, the mechanical properties of 3D printed specimens of different printing parameters were tested under tension. The printing parameters of these specimens were: surface resolution, infill density, and print orientation. Parts were printed in Onyx nylon with a Fused Deposition Modeling (FDM) printer called the Markforged Onyx Pro. Factorial sets of specimens using all various parameters are printed and tested to create a reference table for future engineering projects. Specimens are then printed as composite variations with continuous fibers in order to understand the benefits a composite may have.

A racecar’s suspension is one of the key contributors to its performance on a track. Each component – springs, shocks, links, etc. – can be dealt with as a variable within a mathematical model. There are hundreds of combinations of these variables, with each change affecting the stiffness ratio. Using the sway bar as the variable of interest, data acquisition, and computer modeling, a mathematical was developed for predicting the stiffness ratio as a function of sway bar diameter. This model can simplify the time-consuming iterative process that is “racecar setup” by allowing a race team to plug numbers into an equation to make predictions instead of conducting on-track test sessions to determine the results of each component change.
PDF: Attached to this email.

In our project, image tracking was employed to provide a honing mechanism for a robotic "scorpion tail" attached to a small Remotely Controlled Vehicle. The car will be controlled wirelessly through a web interface, with mobile phones being the target user. Like the Mario Kart Versus Mode, where multiple cars drive and bump into each other, the vehicle will be controlled wirelessly while the "tail" is actively seeking targets and upon close proximity will "pop" the balloon. Each car will have 3-5 balloons to start, and the objective of the tournament will be to hunt down the remaining cars and “pop” their balloons, until all cars lose their balloons and the victor remains with at least one balloon intact. Python and Google Cloud were used to make a server with for the mobile website, and C++ was used to relay the commands sent wirelessly to the vehicle's two DC motors. Image tracking was implemented using the popular computer vision OpenCV library in python. The research will conclude with a tournament on Pi day (March 22, 2019).

In our project, we aimed to design an autonomous rover similar to that of the popular Mars rovers such as Curiosity. Our rover employs a differential drive system with two continuous rotation servo motors that are controlled with the popular ROS robotic programming library in C++ and Python. A navigation algorithm employs the known position of the robot gathered from a magnetic encoder on the motors and the multiple optical range fidners placed around the vehicle to avoid obstacles on route to its destination. A camera is employed to detect target objects for simple pick-and-place tasks using its DC motorized gripper placed at the front of the vehicle. We have successfully built this vehicle and will demonstrate its capabilities at the 2019 IEEE R5 robotics competition in Lafayette, Louisiana as well as at the SRS presentation day.

Rotating Precision Mechanisms, Inc. (RPM) requested that TCU Senior Design update their current Laser Position Accuracy Test Set, which utilizes a laser to calibrate rotating pedestals. RPM positions this test system at a range of distances from a rotating mirror, passes a laser beam through an optical system to the rotating mirror, and measures the offset of the reflected laser dot in order to test the pointing accuracy and repeatability of their positioners. RPM requested that the redesigned test set deliver a reflected laser dot size within 0.125 inches when the test system is any distance between 10 and 100 feet from the rotating mirror. Our prototype for the redesigned Laser Position Accuracy Test Set relies on an optical component called a beam expander to cleanly extend the laser beam at the desired dot size over the specified range of distances. In order to design and manufacture this beam expander, our team researched optical collimators, beam expanders, and lenses in addition to using an Optical Ray Tracing software to model potential beam expander designs. After constructing and testing a working prototype, we completed several iterations in order to improve the resulting laser dot size. Finally, we compared our beam expander design to an Edmund Optics research grade beam expander to further quantify the success of our design.

In our project, a control-theory based algorithm would be employed to develop a small electric vehicle that can self-navigate through an unknown course to arrive at the desired location while avoiding obstacles and walls. This project is an extension of our successful project funded last year, in which we were able to operate a partially autonomous car to run around a location, and generate a virtual map. Our team expects to grant the car full autonomy like a self-driving car and let it travel through a relative abundance of places to create computer models of critical infrastructures without the help of humans. The success of this project will have a broad impact on society. First, this capability would be useful in self-driving cars, which allow drivers to spend their time more productively instead of driving to work or assist disabled people. Second, the car can generate a simulated model of places that help to analyze unknown locations. Finally, the project can surely create a platform for future TCU engineering students to learn about self-driving car technology and machine learning. This project is expected to succeed due to the achievements we gained from the previous project.
The algorithm will be written in Python/ROS, controlled by Raspberry Pi 3, and tested on a walled course constructed by us. It should be able to navigate a course, without having already driven through it. Another special feature is that the car will also precisely arrive at a pre-determined location.

Flatfoot and cavus foot are postural issues that affect approximately 40% of people and can be corrected by means of orthotic inserts for shoes. A digitally reconfigurable mold is being developed as a tool for orthotists to visualize and fabricate orthotic inserts. The surface will be formed by an array of solenoid actuators controlled by the orthotist. The patient will stand on the reconfigurable surface while the orthotist evaluates the patient’s needs by manipulating the surface. Once the orthotist is satisfied with the array, the surface position will be held by a clutch system, so the patient can step off the surface and the surface positions can be recorded. This work describes my development of a prototype mechanical clutch for the digitally reconfigurable surface. The result of this project is a proof-of-concept design of an array of twenty-five physical clutch points which may be individually addressed by means of servo motors controlled by an Arduino microcontroller. With the development of this prototype, it is believed that such a control interface could be implemented on a system large enough for an adult human to stand on. This proof-of-concept is a small step in a larger project of developing a full-scale reconfigurable surface by which an orthotist could create posture correcting devices.

Urban Heat Islands (UHI) describe a phenomenon of increasing ambient temperature in densely built areas of cities as compared to rural areas. Impervious cover, ubiquitous in urban areas, appears to absorb solar radiation and reemit that radiation as heat. Urbanization and UHIs have impacts that range from local to global scales and can be found in cities of all sizes and climatic regions (Fernando 2013). This study focused on Tarrant County, Texas and analyzed changes in impervious surface cover and average monthly temperatures at four different NOAA weather monitoring stations over approximately 60 years in a search for urban heat island effect. Temperature analysis indicates an increase in temperature over the 60-year period. This study aims to determine whether that temperature increase is due to UHI.

The Urban Heat Island effect, or UHI, describes a phenomena involving heighted temperature indices in metropolitan areas when compared to surrounding rural landscapes. An increasingly relevant area of study following the advent of global warming, today, the associated “infernos” plaguing urban landscapes across the globe have the capacity to seep beyond concrete jungle walls, to reap havoc on once lush and vibrant vegetation and ecosystems. The following analysis combines an understanding of the Urban Heat Island effect with Remote Sensing technologies and Landsat Aerial Imagery to uncover the impact of urbanization in Seattle, Washington. With an emphasis on illustrating change over time, historical data surrounding imaging and climate trends further support this GIS analysis of vegetation in Seattle.

Bats are critical to their surrounding environment, thus we need to know what resources bats need to survive, such as water. Many available water resources in urban areas are ephemeral and dry up during the hot Texas summers. We explored bat resource use in an urban environment by radio-tracking bats in local Fort Worth parks, Foster Park and Overton Park. We used Lidar and 3D mapping in ArcGIS 10.6 to portray our study site where bats were tracked. Using digital elevation, we evaluated high elevation points in the parks that can be used in conducting future surveys.

Water resources are critical to areas experiencing urbanization and a rapidly increasing population. The depletion of these resources due to either human usage or environmental factors has the potential to lead to water scarcity in surrounding areas. A GIS analysis was conducted on Lake Mead near Las Vegas to assess the change in water level over time.

Healthcare deserts are an emerging problem in the United States, especially in rural areas. Individuals in these areas do not have access to adequate healthcare, and in most cases they are forced to travel long distances to receive the care they need. In the Permian Basin, this can be of concern for those working in the oil industry as well as their families. A GIS analysis was conducted to identify healthcare deserts in this area.

Monarch butterfly populations in North America have declined by approximately 80% over the last 20 years. Many contributing factors are responsible for this decline, however the loss of Milkweed has been identified as a major factor. Milkweed is the primary food source for Monarch caterpillars. A GIS analysis was performed to identify milkweed resources in the North Texas area.

Coffee is one of the most popular drinks in the world. Americans generate on average 12600 tons (around a thousand school buses) of coffee grounds per day – which is most often landfilled. This typical waste stream represents a potential feedstock for developing carbon-based materials with applications across numerous disciplines. For example, initial research in Dr. Harvey’s Lab on spent coffee grounds, received from Avoca Coffee Roasters in Fort Worth, has shown that charring the grounds at 350 ℃ improved its Lead removal abilities. Our research will build upon this previous research by conducting more controlled studies to evaluate those earlier results and determine the optimal charring temperature for producing charcoals for water filtration applications. Since we had no control over how the previously donated coffee grounds were brewed by Avoca Coffee Roasters, we decided to design an experiment that will give us more control over the brewing process. The coffee beans were obtained from Avoca Coffee Roasters and brewed according to the Golden Ratio. We will also study the surface properties of charred coffee grounds produced at 350 ℃, 450 ℃, and 650 ℃ from Ethiopian and Mexican coffee grounds.

Zebra mussels, Dreissena polymorpha, are an invasive species of freshwater bivalves that have recently spread into bodies of water across North America. Zebra mussels inhabit the shallow waters of lakes and tightly attach to any and all hard surfaces. They are efficient filter feeders and can filter up to 1 L of water per day per mussel. This increases the clarity of water dramatically which alters the lake habitat for other lake species. In this analysis, water clarity data was mapped for “infested” lakes for the years 2008 (pre-zebra mussels), 2010 (one year after zebra mussel infestation), and 2016 (seven years after infestation). The average clarity of the lakes increased by 9.36%, with larger lake clarity increasing the most dramatically.

Habitat loss, disease, and land-use change has led to a sudden decline in bat populations in the US. Thus, there is a need to determine the extent of the impact before we can effectively implement counter-measures. One way to assess the impacts is to monitor areas with a high abundance and species diversity, such as Big Bend National Park with 25 of the U.S.’s 47 native bat species. We therefore assessed whether 1) acoustic monitoring at the park was a feasible technique and 2) if the diversity of species recorded and their activity patterns could contribute to national long-term monitoring.

A GIS and remote sensing analysis of Rwanda was conducted to analyze changes in land cover, urbanization, and croplands over time. Data mapping changes in major crops productivity throughout time was also analyzed and combined with information on elevation and soil conditions. All factors were analyzed to identify the location and suitability of soils for each major crop.

Nitrate contamination of groundwater in the Seymour Aquifer is a well-known issue that has been documented since the 1960's. Concentrations as high as 35 ppm NO3-N have been reported, which is a startling 3.5 times the EPA allowable standard for drinking water. While most water from the Seymour Aquifer is used for agricultural irrigation, a portion is still used for domestic purposes and therefore poses a risk to human health. While this problem may have been recognized, the specific source of this contamination remains unknown. Three potential sources of nitrate within the aquifer are being considered in this study—the geological makeup of the aquifer, the agricultural contribution of nitrate from fertilizers, and the historical land use change of the area above the aquifer.

My research will combine various statistical and geospatial technologies in order to 1) view nitrate contamination as a function of well depth and 2) Determine the temporal change in NO3-N concentrations over a distribution of well depth. Readily available groundwater quality data from the Texas Water Development Board will be used in conjunction with geospatial analysis and statistical analysis to identify well depth distribution and changes in the aquifer's water quality with respect to well depth over time. After a thorough analysis of the site area via the aforementioned methods and technologies, a portrait that depicts the both spatial and temporal changes of nitrate contamination in Texas's Seymour Aquifer ought to be painted.