Exploration of Fabrication Methods for Origami-Inspired Systems to Address Crease Pattern Incompatibility

Origami-inspired structures are theoretically intriguing designs, but complications arise when they are constructed of more substantial mediums because the fold patterns are often incompatible with the material thickness. This project explores fabrication methods to create foldable structures that capitalize on the increased stiffness and flexibility of the system, while addressing the crease pattern incompatibility. Past work at Bucknell identified the Miura-Ori fold as a basis for experimentation. This project explores the feasibility of utilizing fiber reinforced polymers (FRP), a series of rigid plates bonded to a base material with flexible hinges between stiff areas. The materials used for construction experimentation are an epoxy resin and fiberglass cloth due to their inexpensive nature and adaptability. Initial testing involved perfecting the application method of the epoxy on the fiberglass. The critical issue here was ensuring a clean edge on the epoxy to control the location and angles of the hinge. The final application method selected for moving to design was using rubber-based adhesive as a seal on the fiberglass fabric to outline the geometric pattern. A double-sided, wet layup application method of both the rubber-based sealant and the epoxy resin was deemed most effective at ensuring consistent thickness and clean edges. Using this technique, a successful 3D model of the classic Miura-Ori fold has been constructed and analyzed for necessary improvements. Ongoing work focuses on improving the system’s ability to maintain its form after folding and using the same materials and fabrication to scale-up designs and work towards real-world applications.

Long-Term Web Accessibility Evaluation for Small Businesses in Union, Snyder, and Northumberland Counties

The purpose of this study is to evaluate and improve upon the long term sustainability of web accessibility initiatives for small businesses in Union County, Snyder, and Northumberland counties. By implementing a series of web accessibility workshops, this research focuses on instilling fundamental accessibility features such as color contrast and alternative text among participating businesses. These interventions are designed to enhance the digital inclusivity of small enterprises by equipping them with the necessary tools and knowledge to implement and maintain accessible web practices. The evaluation of these businesses will occur longitudinally, providing insights into their ongoing capacity to uphold accessibility standards. This study aims to address the critical need to reduce accessibility barriers for small businesses in rural areas. Consequently, the anticipated impacts include the empowerment of rural businesses through increased digital presence and engagement, and the fostering of a more inclusive digital environment, which contributes to the diversification of local industries and enhances overall community resilience in the digital age.

Crawling gait design for a caterpillar-inspired robot with tendon-driven flexible bending segments
Soft crawling robots can potentially access locations that are unreachable by humans and traditional rigid robots, playing a crucial role in conducting missions like environmental monitoring or search and rescue. Due to their flexible body structures, soft robots can adapt to uncertain environments and operate safely in contact with humans. We present a new design for a caterpillar-inspired soft robot in the form of a series of 3D printed flexible tendon-driven bending segments with individual motor control. A constant-curvature quasi-static kinematic model of the robot locomotion is developed, and we describe periodic gait inputs that coordinate bending in multiple segments to lift the prolegs and move forward in a traveling-wave motion. We present simulations and experimental results for locomotion in a straight line, along with experimental demonstration of a modified segment design with an additional degree of freedom for steering navigation.

Development of Soft Gripper and Body for Caterpillar-inspired Robot

This project is designing a caterpillar-inspired robot that can climb on pipes or branches. The project involved creating 3D models of clamps and brackets using Solidworks and Ultimaker Cura and printing hard and soft parts using PLA and Filaflex respectively. The force sensors are connected to the Arduino board to collect force data, which is analyzed using MATLAB. The N20 motor is used to activate and reset the grippers and pull up rigid parts of the body to deform. The design process involved exploring different gripper and body designs, including those with flat and curved stems as well as fin ray designs. Finally, data were collected to evaluate the force ratio metrics and predict the movement of the body. We aim to create a gripper with the highest force ratio and coordinate the body with the gripper to help to robot climb on the pipes. In the experiment, we measured the four forces of each gripper and find the parameters that have the most positive effect on gripping and extracting force and the most negative effect on less closing and opening force which helps the grippers work much better helping caterpillar robots to move, climb and hang. The force required to pull up the body was also measured and analyzed with position.

Vehicle alignment measured and adjusted by wheel forces on steer-by-wire systems

To optimize fuel efficiency and tire wear, all modern automotive vehicles have service intervals after which the front wheels are re-aligned. An offset alignment causes unwanted forces and moments on the wheels and steering system that reduce fuel efficiency and increase tire wear. An alignment procedure is normally done using a machine that can measure the angle of wheels, which are adjusted using mechanical linkages on the vehicle. New electronic steering systems, such as steer-by-wire, offer a new alternative to this angular alignment procedure. Since these steering systems can control the angle of wheels independent from the steering wheel position, the alignment of the wheels can be done without adjusting mechanical links. By measuring forces on the wheels with onboard sensors, a relationship can be developed between them and the correct alignment angle of the wheel. A vehicle model was created and tested against an electric research vehicle equipped with a steer-by-wire system. The research vehicle was equipped with a set of wheel force transducers that were able to measure the forces and moments on the front wheels in all directions. It was found that adjusting the wheel angle to minimize both the rolling resistance and slip angle produced the highest increase in fuel efficiency and the largest decrease in tire wear. Additionally, this vehicle model serves as a bench block for an active alignment algorithm that can take real-time wheel force measurements and automatically adjust the steering systems based on these loads.

Building Large Language Models

ChatGPT became available in November 2022. It is the most widely used large language model. This research explores uses of ChatGPT in a variety of industries. We also explore building our own large language model using nanoGPT. NanoGPT offers the opportunity to create a small scale large language model on a single computer. Research questions are ‘How are industries using ChatGPT?’ and ‘How can we create our own large language model outside of ChatGPT?’

Machine Learning to Detect Damage in Non-Linear Systems

Structural and mechanical systems wear and fatigue over time and managing these effects has proven to be a critical challenge for engineers. Monitoring damage in systems allows time, resources, and money to be properly allocated to repair or replacement. Many methods exist for damage detection in linear systems, but nonlinear systems have few methods available. Sensitivity vector fields (SVFs) are one way of detecting damage in nonlinear systems. SVFs quantify how nearby dynamic paths in phase space, called trajectories, separate due to damage. The magnitude of a sensitivity vector is expected to correlate to the amount of damage while the direction of a sensitivity vector can indicate the type of damage. This research used machine learning (ML) to expand SVF damage detection into the multi-parameter domain and to test its effectiveness on experimental data sets. Experimental data from a Chua circuit was processed in MATLAB and then the damage was classified using a stacked meta-learning model. A mix of support vector machines and random tree ensembles provided the best classification and regression model. The model was able to predict changes in all three system parameters with an MSE of less than 2%. Simulated data was created for a five-body mass-spring system driven by a chaotic signal. Accurate classification and regression results were obtained for limited damage scenarios, such as multi-level damage at one location or a fixed amount of damage at multiple locations.

Using AI in Assistive Technology to Support Effective Note-Taking Skills

In the ever-evolving landscape of engineering education, we are at a point where traditional instruction can be supplemented by AI to bridge the gap between current and best practices in the area of universal access. We have analyzed the Explicit Instruction Model, which matches well with STEM teaching approaches in higher education, with a Universal Design for Learning (UDL) lens to identify points in the instruction process where AI could support learning. From this activity, we determined that AI can facilitate learning through summarizing lecture content, identifying main ideas, developing checks for understanding, and providing supplementary resources for students. Once these features were identified, their efficacy was tested on three prominent generative AI platforms: ChatGPT 3.5, Gemini, and Bing AI. We created a prototype that analyzes a video of a STEM lecture and extracts a text transcript. The transcript data is then sent to the AI platform, which returns summaries, main ideas, checks for understanding, and supplementary resources pertaining to the lecture.

We are investigating more AI-powered tools, focusing on their ability to provide real-time feedback and adapt to individual student needs. Additionally, various metrics and assessment strategies are under consideration to allow educators to quantify the impact of AI on student engagement and learning. In the future, we plan to deploy the AI prototype to collect Bucknell student feedback. By drawing a clear connection between explicit instruction models, UDL, and AI, engineering instructors can create more effective learning experiences for their students.

Chaotic Synchronization and Damage Detection

The general approach to the research was to attempt to build as much knowledge on chaotic systems and synchronization as possible in order to explore potential avenues for future research. Seeing as I was unfamiliar with the code and I needed to simulate the relevant chaotic systems, an introductory course to that syntax was one of the initial aims of the research. Familiarizing myself with differential equation calculations as well as various ways of displaying and interpreting the relevant data were essential skills I picked up in the wee stages of this research.
A literature review of relevant papers followed the introductory phase of the research where I looked for, read, and summarized salient aspects of academic articles in relation to chaotic synchronization and damage detection. Now armed with relevant information, skills, and equations I began to replicate previous works cited in articles in order to reconcile all the various skills I had annexed over the weeks and cement my understanding of them as a single analytical procedure.
I mostly worked in MATLAB and relied on the advanced integral calculator ODE45 to run most of my simulations. By simulating appropriately coupled chaotic systems I attempted to see if I could determine the difference between parameter values in the equations of the two involved systems (damage). I did this by evaluating one of the output component’s clearly defined properties. The result was that I found two ways of reliably determining the value of unknown parameters in the driving equation by comparing the output values:

1)The value of the deviation in the first peak or trough from the start line
2)The average of one of the predetermined outputs of the system
While the results varied for different chaotic systems, in most cases one or the other or both gave a good estimation of the value.

Aside from these findings I learned a few new math-oriented skills like how to calculate eigenvalues and determine the nature of a chaotic system based off it, how to write a state space equation and how to plot vector fields in MATLAB.

Energy Harvesting for Residential Microgrid Distributed Sensor Systems

Microgrids are localized, independent power grids that can operate while connected to the larger electrical grid. These systems make intelligent decisions regarding power management and use an array of components to monitor power generation, consumption, and environmental conditions. While this technology can save money, complex installation and maintenance has limited the adoption of microgrids in residential spaces. To simplify this technology, the next evolution of microgrid components includes sensors that are wireless and ambiently powered. Energy harvesting circuits can be incorporated into microgrid sensors, enabling them to recapture otherwise wasted environmental energy. Equipping microgrid sensors with energy harvesters simplifies the end user experience by eliminating the need for cable routing. Implementing energy harvesting techniques results in a microgrid that is easier to deploy, cleaner, and requires less maintenance. In this poster, we share the design, implementation, and evaluation of a hybrid energy harvesting and battery-powered sensor. We assessed the sensor’s performance through experiments, demonstrating its potential to operate autonomously and reliably for a decade or more without maintenance.