Integrating Research into Capstone Projects: A Case Study of Dual-Instructor Approach

Introducing research experiences to undergraduates presents both structural and pedagogical challenges. Traditional computing curricula emphasize implementation and correctness, often at the expense of open-ended inquiry. Although capstone project courses offer opportunities for applied learning, these projects are typically narrowly scoped, rely on curated datasets, and follow prescriptive development paths. As a result, students rarely encounter the uncertainty, iteration, and problem-framing central to real-world research. This study explores how undergraduate research skills can be cultivated through a redesigned senior-level computing capstone project course. A dual-instructor teaching model was employed to separate course logistics from deep, research-oriented mentorship. One instructor supervised learning objectives, scheduling, grading, and coordination, while the other guided a selected team through weekly mentoring-mirroring a supervisor-client dynamic. This structure enabled technical exploration within a scaffolded educational environment. In the pilot case, a student team developed a seizure-like motion detection system using wearable IMU sensors and a deep learning model. The team assembled hardware, collected and labeled motion data, and constructed a full machine learning pipeline including CNN-LSTM modeling and evaluation. Students engaged directly with sensor data, bridging theory and application while navigating the complexities of empirical research. The experience led to improved technical fluency, increased confidence in tackling ambiguous problems, and stronger research habits of mind. This model lays the groundwork for an instructor-anchored multi-mentor structure capable of scaling research experiences across multiple teams within a capstone course.