REU 2024 Summer Projects
List of Summer Projects
- Development of AR and VR Aircraft Engines for Aviation and Engineering Programs
- Network Conversation Chains
- Fuel Cell - A Clean and Efficient Energy Technology
- Airworthiness Standards and Requirements for Certifying Autonomous Aircraft
- Design of Advanced Control Algorithms for Robot Manipulator Joints
- Saving Three Birds with One Stone: Collision Avoidance Between Aircraft, Noncooperative Aircraft, and Birds using Artificial Intelligence
- Robotic Orthosis to Maintain Astronaut Strength in Space
- Byzantine-Resilient Multi-UAV System
Summer Project Descriptions
Development of AR and VR Aircraft Engines for Aviation and Engineering Programs
Dr. Stephanie Fussell and Dr. Benjamin Kwasa
There are many virtual and augmented reality (VR, AR) maintenance programs that have been developed for military purposes. A smaller number have been created for commercial flight operations, generally for Part 121 air carriers (e.g., airliner). Few have been created for students: those pursuing training as aviation maintenance technicians (AMTs), engineering students, pilots who must understand the systems of the aircraft they will operate, and others who will maintain aircraft for airlines, flight schools, and other commercial enterprises.
The REU students participating in this project will develop AR and/or VR engines using Unity, Unreal, or an AR authoring program (i.e., Scope AR). They will design the virtual engines using 3D models, develop the models for the different virtual environments, and test the functionality and usability of the virtual engines in both environments.
The students will learn about accessibility in virtual environments while they design the programs. The AR engines will be tested in different physical environments and ambiance (e.g., dim lighting, crowded spaces). The VR engines will be tested in different headsets that have different resolutions, framerate, software and hardware, etc. Additionally, the students will learn how to program anchored and non-anchored textual information, visual and aural cuing, and simulated technical faults and other learning elements that may be added or removed. Finally, the students will learn how to assess the programs using systems engineering, human factors, and user centered design standards and best practices.
Network Conversation Chains
This project aims to develop a framework for analyzing "conversation chains" in network packet traces (PCAP files) within an autonomous system's communications network. This analysis will help assess network performance, functionality, and security vulnerabilities.
A "Conversation" refers to a complete sequence of request/response packets exchanged between two hosts. For TCP, this includes the entire content stream of the TCP connection. For session-less protocols, it involves packet exchanges as defined by the application, like an ARP request paired with its reply, or a DNS request and its server responses. Conversations are identified based on protocol-specific header fields. In a slightly more realistic situation, we can have a chain of an HTTP request, which then results in more DNS lookups and data requests (Figure 1).
This ends up with our chain looking much more like a tree (Figure 2). Once these trees are isolated from the rest of the packet data, we can open up a wide range of analysis opportunities, allowing for unique insight into the data for other research topics.
Fuel Cell - A Clean and Efficient Energy Technology
Our project focuses on fuel cells, the most energy efficient energy technology. Fuel cells have a wide range of applications, such as drones, cars, trains, airplanes, and power stations. Basically, anywhere you need electricity, you could use a fuel cell. Fuel cell technology can significantly reduce greenhouse gas emissions while providing long lasting power.
The major goal of this project is to evaluate and compare various fuel cell technologies. Students will learn the principles of fuel cell operation, various fuel cell technologies and fuel cell applications. Previously we demonstrated a hybrid fuel cell-battery power for drones, extended the fly time from 30 minutes to nearly two hours.
As an undergraduate researcher participating in this project, you will be trained in research methods and cutting-edge fuel cell technologies and have the opportunity to incorporate fuel cells to power an autonomous system, such as a drone. You will gain hands-on experience on fuel cell operations. My research group includes faculty, postdocs, graduate, and undergraduate students. I welcome passionate talent to join my group.
Airworthiness Standards and Requirements for Certifying Autonomous Aircraft
Dr. Jason Lorenzon and Dr. Syed Shihab
Numerous aviation manufacturing firms are actively engaged in the development of fully autonomous aircraft, which promises to be safer, and more intelligent and efficient compared to human-piloted aircraft. Such autonomous aircraft are envisioned to provide package delivery and air taxi services to urban areas, as depicted in Figure 1. Although test flights have successfully demonstrated the capabilities and effectiveness of autonomous aircraft, their integration into national airspace faces a major hurdle—the certification process by the Federal Aviation Administration (FAA) in the United States.
The existing certification regulatory framework lacks clear airworthiness standards and requirements for certifying aircraft equipped with autonomous flight control systems and artificial intelligence (AI) technology. This project's primary objective is to address this crucial gap in aviation regulations by formulating performance-based airworthiness standards tailored for autonomous aircraft and their onboard AI technology. The research will determine the types of performance data that need to be collected for certification of autonomous aircraft. Additionally, the study will model and quantify safety risks associated with the performance of autonomous aircraft. The research will focus on autonomous electric Vertical Take-Off and Landing (eVTOL) aircraft and Unmanned Aerial Vehicles (UAVs) as case studies to examine their performance characteristics.
The student participating in this project will expand their knowledge of the challenges associated with certifying autonomous aircraft and aircraft employing AI technologies. Through this project, they will have the opportunity to contribute towards developing airworthiness standards and requirements for certifying autonomous aircraft with desired safety assurances.
Design of Advanced Control Algorithms for Robot Manipulator Joints
This project presents a valuable opportunity for in-depth research experience in areas such as joint control, kinematics, path planning, statics, and dynamics. Participants will gain a comprehensive understanding of the intricacies of robotic systems, preparing them for research and education in the field of robotics.
The aim of this project is to develop advanced control algorithms, such as a model predictive controller, to manage the movement of robotic manipulator joints, such as the one shown in Figure 1. Model Predictive Control (MPC) is a control strategy that employs a system model to anticipate future outcomes, optimizing control actions over a predetermined time horizon. This approach is continually refined with real-time data, making it highly effective for complex, multivariable processes with various constraints.
Central to this REU project is a 4-DOF (Degree-Of-Freedom) serial robotic manipulator equipped with a tendon-based two-stage gripper and an RGBD camera. The manipulator is designed with four motors located at key joints: the base, shoulder, elbow, and wrist. Operationally, it is directly controlled using MATLAB and Simulink, ensuring seamless integration and ease of use for both research and educational applications.
Saving Three Birds with One Stone: Collision Avoidance Between Aircraft, Noncooperative Aircraft, and Birds using Artificial Intelligence
Emerging Advanced Air Mobility (AAM) aircraft – such as drones or uncrewed aerial vehicle (UAV) and electric vertical takeoff and landing aircraft (eVTOL) – are expected to make urban air taxi services and package delivery by air an everyday reality soon. These aircraft will carry out their operations at the lower altitudes of the National Airspace System (NAS), where they will share the airspace with birds, as shown in Figure 1, as well as other aircraft, which may be noncooperative. Bird strike risk probability with aircraft is highest at these low altitudes, with most bird strikes happening below 1000 m. Also, the risk probability of collisions of AAM aircraft with noncooperative AAM aircraft is high in these low altitudes. So, the risk probability of collisions of AAM aircraft with both birds and noncooperative AAM aircraft need to be minimized and considered during flight planning to ensure safe collision-free AAM operations in the NAS, as illustrated in Figure 2.
This project aims to address these critical research needs through the following steps: 1) modeling and predicting the flight tracks of birds and noncooperative AAM aircraft using an AI-based model; 2) differentiating between birds and AAM aircraft based on their flight tracks using an AI-based classifier; and 3) feeding outputs from models of Steps 1 and 2 to an aircraft flight trajectory planner. All models and solutions will be rigorously evaluated using validation and test data and numerical simulations and benchmarked against traditional methods found in the literature.
The REU students participating in this project will develop skills on machine learning/artificial intelligence (ML/AI), data analysis and safe flight trajectory planning for aircraft.
Robotic Orthosis to Maintain Astronaut Strength in Space
Astronaut exposure to the microgravity environment of spaceflight results in a loss of muscle mass and a decline in muscle strength and physical endurance. Astronauts need about two-hours of exercise each day to alleviate this loss of muscle mass. Existing remediation systems, such as whole-body exoskeletons, are too bulky and complicated given the physical constraints in space travel.
This project is dedicated to developing a compact, universal and reconfigurable wearable robotic orthosis that can be used for any joint for astronauts depending on their individual health needs. Students will receive research training on this robotic project. Specific learning opportunities may include but are not limited to meta-analysis, mechanical design skills, electronic control and sensing methods, experimental setting and data analysis approaches.
Byzantine-Resilient Multi-UAV System
Dr. Tracy Chen and Dr. Rui Liu
This project focuses on the security issues in multi-agent reinforcement learning (MARL) on multiple, cooperative UAV systems. Our focus is addressing Byzantine attacks, which can significantly disrupt learning performance, leading to mission failure and unreliability. To tackle this problem, we have developed a new Byzantine-resilient MARL algorithm, which ensures our UAVs can maintain accurate and reliable learning in the presence of attacks. We have performed software simulations in two multi-UAV navigation scenarios as shown in Figure 1. The results are very promising.
The objective of this project is to deploy this proposed algorithm in real drones in the lab environment, simulating various multi-UAV navigation scenarios while executing realistic Byzantine attacks. This hands-on approach will allow us to evaluate the algorithm's performance under real-world conditions by closely observing how it responds to real-world cyberattacks and how it responds to dynamic environments and other unforeseen challenges.
