K-12 Research Project

MIN Lab offers high school students an experience on the data science using artificial intelligence and machine learning. This will provide basic knowledge on emerging data science using AI/ML and its application to real world problems. Upon completion of projects, the certificate will be issued. The project consists of multiple steps as below. Students can stop at the completion of Step 1 or proceed to Step 2 and 3 if interested.

1. Step 1: Data Analysis using Statistical Approach and AI/ML: Useful for data analysis in engineering, science, economics, etc.
2. Step 2: Real industrial problem solving or practical problems: Hands-on experience on IOT devices, coding, etc.
3. Step 3: Research engagement

Data Analysis using Statistical Approach and AI/ML

• Aim: To study the influence of 3 different input parameters on the cutting and thrust force during ultra-precision machining of sapphire 
• Phase 1: Use of statistical experiential design – the data has already been collected and summarized. You will be performing the analysis
• Phase 2: Use of data science approaches to study the same system – data has been collected but it needs to be organized prior to performing the analysis
• Phase 3: Writing a brief report on the findings
• Reading materials and project data (password required)

To apply for this program, please contact sangkee.min at wisc.edu.

Notice: Due to increasing number of students applying for this program, we are looking for donation to cover materials, consumables, and graduate students’ time. Any amount would be appreciated. Please contact us for the detailed information.

Project Title: Enhancing Industrial Safety: Integrating Body-Worn Sensors and Camera Monitoring for Human Action Recognition

High School Project Leader: Aiden D. Park, Dougherty Valley High School, San Ramon, CA

Project Description:

Empowering Industrial Safety through Advanced Deep Learning and Machine Learning Techniques: A Comparative Study of Human Action Detection Using Body-Worn IMU Sensors and Camera Data. Our research delves into the detection of various actions by human operators within manufacturing industries, comparing the effectiveness of IMU sensor data and camera data (RGB and Depth). Our ultimate goal is to predict human actions proactively, thereby preempting safety incidents before they occur.

Work performed by the student (Needs some finetuning after discussing with the student):

1. Start by being able to parse and understand the IMU sensor data to better detect and classify different human actions.
2. Compare the IMU sensor information with RGB camera information for a similar activity to better understand the computation requirements and model performance.
3. Learn about different feature extraction techniques.
4. Compare different machine learning algorithms for human activity detection.
5. Identify how the extracted features contribute to the model’s performance.
6. (Optional) Using LLMs or attention-based network to predict safety incidents from both IMU and Camera data (IP Cameras)