Past Research

Mission: Enable the curation of a database for various machine parameters (e.g., current, vibration, temperature) by developing a unique, low-cost, non-invasive, scalable solution. This solution aims to help makerspaces develop actionable insights, leading to smoother operations and maintenance. To achieve this, the solution must be able to be handled by anyone, regardless of background, standalone, and have clear documentation. This will allow individuals of any background to create various algorithms for machine uptime, predictive maintenance, or anomaly detection. Read more here.

Current Iteration: The current setup uses a Raspberry Pi connected to a Teensy 4.0, which reads inputs from an MPU6050 and an MLX90614 sensor, then transmits the data via AWS for future reference.

The Makerspace Accessibility Group is dedicated to designing inclusive and adaptable makerspace experiences that encourage participation from nontraditional users—specifically students from majors and disciplines that do not typically engage with makerspace tools. Our research aims to identify barriers to access, provide solutions for a more inclusive makerspace culture, and create an environment that fosters creativity and innovation for a broader range of academic backgrounds. To better understand the circumstances of many students and possible reasons for biases, our research group has collected a library of applicable articles, past research, and interviews and combed through the accumulated data. We have also run multiple pop-up and workshop events to better understand student interactions and beliefs surrounding the makerspaces on campus. Read more here.

Our research focuses on collecting and analyzing acoustic data to identify operational anomalies in bandsaw performance. Building on previous studies that confirmed the effectiveness of sound data in characterizing CNC machine performance, we are adapting these findings to study bandsaw operation through real-time anomaly detection. Specifically, we’re investigating the use of refined FFT (Fast Fourier Transform) frame sizes to capture and interpret subtle acoustic signatures, which reveal critical information about operational stress and potentially incorrect settings as they occur. Read more here.

Our team tackled improving safety monitoring in busy makerspaces. Recognizing that constant human supervision of equipment is difficult, we focused on automating safety detection for the bandsaw. We defined “safe” bandsaw usage as keeping hands away from the blade and using appropriate cutting speeds. Using a YOLOv5 model, we trained a computer vision system on nearly 1000 frames of video showing both safe and unsafe bandsaw operation. Our model achieved ~90% accuracy in tracking the user’s hands and the bandsaw blade. We then developed an Android app that uses TensorFlow Lite to provide real-time alerts via the phone’s flashlight if unsafe usage is detected. This project won first place for Software and Applications at the 6th Annual VIP Innovation Competition in Fall 2022. Read more here.

Our team developed a people counter app for the Invention Studio makerspace to provide insights into usage patterns. The app leverages Amazon Rekognition, a computer vision service, to automatically detect and count individuals in images captured by a phone’s camera. This data is then stored in a local database along with timestamps, allowing for analysis of usage trends over time and identification of high-traffic areas. Our initial alpha version successfully demonstrated core functionality, including image capture, people detection, data storage, and visualization through graphs.