Computer Engineering Commons^™

Flight Simulator Modeling Using Recurrent Neural Networks, Nickolas Sabatini, Andreas Natsis

Undergraduate Research & Mentoring Program

Recurrent neural networks (RNNs) are a form of machine learning used to predict future values. This project uses RNNs tor predict future values for a flight simulator. Coded in Python using the Keras library, the model demonstrates training loss and validation loss, referring to the error when training the model.

From Inductive To Deductive Learning, Mikhail Mayers, Brian Henson

Undergraduate Research & Mentoring Program

Using Machine Vision as a way to give information to Prolog. Using Prolog to solve deductive problems and analogical problems without having to manually enter all facts and information.

3d-Printed Leg Design And Modification For Improved Support On A Quadruped Robot, Jasmin S. Collins

Undergraduate Research & Mentoring Program

The Agile and Adaptive Robotics Lab aims to uncover the biological and physiological complexities in animal agility and adaptive control, which can be replicated through robotics and provide further applications in biology and medicine. One project within the lab focuses on understanding structure, actuation, and control through the modeling of a canine quadruped robot.

The AARL has developed a full-body quadruped robot with artificial muscles that control limb movement and a body that is built from 3D-printed parts. This specific project involved modification of these existing parts to (a) minimize deflections in the front legs, causing unwanted lateral and abduction/adduction …

Creating A 3d Printed Bipedal Robot’S Ankle And Foot With Human-Like Motion, Tylise E. Fitzgerald

Undergraduate Research & Mentoring Program

Humanoid robots are being created to replace humans in dangerous situations, assist overworked humans, and improve our quality of life by completing chores. However, current bipedal robots haven’t matched the performance of humans and are still impractical for commercial use.

One of the Agile and Adaptive Robotics Lab’s goals is to create a humanoid robot whose anatomy is similar to the human body. If this can be accomplished, we can have a functioning model of the human body that we can adjust to improve both humanoid robots’ functions and the functionality of our own human bodies. This specific project looks …

Material Parameter Estimation Of Thin Wafers With Terahertz Time-Domain Spectroscopy, Kirk R. Jungles

Undergraduate Research & Mentoring Program

Terahertz Time Domain Spectroscopy(THz TDS) is a spectroscopic technique that can be implemented to perform non destructive material parameter extraction on a variety of materials. Accuracy of these material parameters is often limited by statistical variation between measurements and insufficient knowledge of the thickness of the slabs being measured.

The goal of this project was to develop an in house procedure that would allow us to perform THz TDS on thin wafers using an up to date signal processing algorithm that would provide accurate predictions for the thickness of the wafers, reliable estimations of the wafer’s material parameters, and demonstration …

Omni-Gravity Hydroponics System For Spacecraft, Tara M. Prevo

Undergraduate Research & Mentoring Program

Effective omni-gravity hydroponics will allow astronauts to supplement nutrition and further close the life cycle of water in orbit, lunar, and Martian conditions. This project determines the operational limits of the test cells for the Plant Water Management Hydroponics mission. A scaled 1-g channel was designed by Rihana Mungin to mimic full-scale performance in microgravity that could be tested terrestrially. This project sought to find the limits of operation of the 1-g test cells and identify failure modes that could pose a safety risk in space. The cells were filled at increments of 20% and cycled from 0.184 to 8.33 …

Simulation Of Human Balance Control Using An Inverted Pendulum Model, Joshua E. Caneer

Undergraduate Research & Mentoring Program

The nervous system that human beings use to control balance is remarkably adaptable to a wide variety of environments and conditions. This neural system is likely a combination of many inputs and feedback control loops working together. The ability to emulate this system of balance could be of great value in understanding and developing solutions to proprioceptive disorders and other diseases that affect the human balance control system. Additionally, the process of emulating the human balance system may also have widespread applications to the locomotion capabilities of many types of robots, in both bipedal and non-bipedal configurations.

The goal of …

The Applications Of Grid Cells In Computer Vision, Keaton Kraiger

Undergraduate Research & Mentoring Program

In this study we present a novel method for position and scale invariant object representation based on a biologically-inspired framework. Grid cells are neurons in the entorhinal cortex whose multiple firing locations form a periodic triangular array, tiling the surface of an animal’s environment. We propose a model for simple object representation that maintains position and scale invariance, in which grid maps capture the fundamental structure and features of an object. The model provides a mechanism for identifying feature locations in a Cartesian plane and vectors between object features encoded by grid cells. It is shown that key object features …

Exploring And Expanding The One-Pixel Attack, Umairullah Khan, Walt Woods

Undergraduate Research & Mentoring Program

In machine learning research, adversarial examples are normal inputs to a classifier that have been specifically perturbed to cause the model to misclassify the input. These perturbations rarely affect the human readability of an input, even though the model’s output is drastically different. Recent work has demonstrated that image-classifying deep neural networks (DNNs) can be reliably fooled with the modification of a single pixel in the input image, without knowledge of a DNN’s internal parameters. This “one-pixel attack” utilizes an iterative evolutionary optimizer known as differential evolution (DE) to find the most effective pixel to perturb, via the evaluation of …

Combining Algorithms For More General Ai, Mark Robert Musil

Undergraduate Research & Mentoring Program

Two decades since the first convolutional neural network was introduced the AI sub-domains of classification, regression and prediction still rely heavily on a few ML architectures despite their flaws of being hungry for data, time, and high-end hardware while still lacking generality. In order to achieve more general intelligence that can perform one-shot learning, create internal representations, and recognize subtle patterns it is necessary to look for new ML system frameworks. Research on the interface between neuroscience and computational statistics/machine learning has suggested that combined algorithms may increase AI robustness in the same way that separate brain regions specialize. In …

An Exploration Of Software Defined Radio And Gnu Radio Companion For Use In Drone-To-Drone Communication, Amanda K. H. Voegtlin

Undergraduate Research & Mentoring Program

In a world that increasingly relies on automation and intelligent robotics, there is a need for drones to expand their independence and adaptability in navigating their environments. One approach to this problem is the use of wireless communication between units in order to coordinate their sensor data and build real-time maps of the environments they are navigating. However, especially indoors, relying on a fixed transmission tower to provide data to the units faces connectivity challenges.

The purpose of this research was to determine the fitness of an on-drone assembly that uses the the NI B200mini software-defined radio board and Gnu …

Automating Knife-Edge Method Of Thz Beam Characterization, Christopher Charles Faber

Undergraduate Research & Mentoring Program

The goal of this project is to create a time and cost-effective solution for THz beam profiling.

The knife edge method of beam characterization is a technique to verify the intensity profile of a beam involving traveling a blade orthogonal to the beam path and measuring transmission in successive steps. We use a vector network analyzer (VNA) to measure S21 transmission from a THz source. Manual implementation of this method was time-consuming and inefficient.

Project hardware includes an Arduino, a motor shield, and a ball screw linear rail with stepper motor actuator. Software was created in LabView and data is …

Binder Free Graphene Hybridized Fe3o4 Nanoparticles For Supercapacitor Applications, Nathan D. Jansen

Undergraduate Research & Mentoring Program

In a world with increasing energy demands, the need for safe and mobile energy storage grows. There are a number of renewable energy sources that can be harvested, however peak demand and peak production times tend to not overlap. As the capabilities of collecting the energy grows so does the need to store the energy for later consumption. The two promising methods of storing energy are batteries or supercapacitors. Both technologies employ an electrode consisting of an active material bound to a current collector. This material participates in a redox reaction, storing charge electrochemically to later be used as energy, …

Laser-Scribed Graphene Micro-Supercapacitors, Kimi D. Owens

Undergraduate Research & Mentoring Program

M. F. El-Kady and R. B. Kaner, “Scalable fabrication of high-power graphene micro-supercapacitors for flexible and on-chip energy storage,” Nature Communications, vol. 4, p. 1475, Feb. 2013.

Supercapacitors are electrical components that have higher energy density than regular capacitors. Currently, they are large and bulky which makes it hard to be implemented into smaller electronic devices or on-chip. In Scalable Fabrication of High-power Graphene Micro-supercapacitors for Flexible and On-chip Energy Storage, El-Kady and Kaner developed an inexpensive and reliable method for scaling down supercapacitors to be approximately 7.53 x 5.35 mm. To make the laser-scribed graphene (LSG) micro-supercapacitors, an aqueous …

Learning In Bio-Molecular Computing Systems, Lauren Braun

Undergraduate Research & Mentoring Program

Many potential applications of biochemical computers involve the detection of highly adaptable and dynamic chemical systems, such as emerging pathogens. Current technology is expensive to develop and unique to each application, thus causing limitations in accessibility. In order to make this type of computing a realistic solution to problems in the medical field, a biochemical computer would need to be adaptable to work in a variety of applications. Banda et al. (2014) previously proposed a first dynamic biochemical system that was capable of autonomous learning. For this project we studied a framework similar to Banda’s but in two separate pieces, …

Improvement Of 802.11 Protocol On Fully Programmable Wireless Radio, Eunji Lee

Undergraduate Research & Mentoring Program

The growth in the number of connected device usage has led to a rapidly increased data traffic on wireless network and the demand for access to high speed and stable Internet connection is becoming more prominent. However, current off the shelf wireless cards are not programmable or observable across layers of the standard protocol stack, which leads to poor practical performance. Thus, Wireless Open Access Research Platform (WARP), a scalable wireless platform providing programmable functionality at every layer of the network stack, has been used for the real-time implementation and improvement of 802.11 protocol.

An Analysis Of Lora Low Power Technology And Its Applications, Gomathy Venkata Krishnan

Undergraduate Research & Mentoring Program

The number of Internet of Things (IoT) devices has exponentially increased in the last decade. With the increase in these devices, there is a necessity to effectively connect and control these devices remotely. Cellular technologies cannot handle this demand since they are not cost effective and easy to deploy. This is where LoRa technology comes handy. LoRa is long-range, low-power, low cost technology that supports internet of things applications. LoRa has many advantages in terms of capacity, mobility, battery lifetime and cost. It uses the unlicensed 915MHz ISM band and can be easily deployed.

This research is focused on setting …

Real-Time Object Detection And Tracking On Drones, Tu Le

Undergraduate Research & Mentoring Program

Unmanned aerial vehicles, also known as drones, have been more and more widely used in recent decades because of their mobility. They appear in many applications such as farming, search and rescue, entertainment, military, and so on. Such high demands for drones lead to the need of developments in drone technologies. Next generations of commercial and military drones are expected to be aware of surrounding objects while flying autonomously in different terrains and conditions. One of the biggest challenges to drone automation is the ability to detect and track objects of interest in real-time. While there are many robust machine …

Generating Adversarial Attacks For Sparse Neural Networks, Jack H. Chen, Walt Woods

Undergraduate Research & Mentoring Program

Neural networks provide state-of-the-art accuracy for image classification tasks. However traditional networks are highly susceptible to imperceivable perturbations to their inputs known as adversarial attacks that drastically change the resulting output. The magnitude of these perturbations can be measured as Mean Squared Error (MSE). We use genetic algorithms to produce black-box adversarial attacks and examine MSE on state-of-the-art networks. This method generates an attack that converts 90% confidence on a correct class to 50% confidence of a targeted, incorrect class after 2000 epochs. We will generate and examine attacks and their MSE against several sparse neural networks. We theorize that …

Emerging Adaptive Architectures For Biomolecular Computation, Matthew Fleetwood

Undergraduate Research & Mentoring Program

The goal of this work is to explore applications of reservoir computing in biomolecular computation. Reservoir computing is a unique model for representing a mapping from one instance in time to a specific output. A neural network of randomly connected neurons is linked with a single output neuron or multiple output neurons. The output neurons are capable of mapping inputs to desired outputs using adaptable algorithms. This framework is investigated by using the Python programming language and object oriented design and programming. Neurons are created in programs by bundling information like input data and attributes of the network, which utilize …

High-Performance Computing For Drought Prediction, Henry Cooney

Undergraduate Research & Mentoring Program

In recent decades, there has been considerable interest in using satellite soil moisture data to examine the global water-energy cycle and manage water resources. Current satellites are limited in their sensing depth, and can only directly measure top soil layers. Using a particle filter, this data may be fused with the output of a hydrologic simulation to improve simulation results, and characterize a hydrologic system at the watershed level. However, this approach increases computational requirements dramatically, and requires rethinking to accommodate data scaling and achieve good performance.

We present a detailed performance study of several alternative implementations of the hybrid …