2026 ASEE Annual Conference & Exposition

Plan of submission for full paper: We will submit a full paper that details the development of the proposed framework, including the functions and costs of major components, photographs of the laboratory and testing environments, numerical experimental results with corresponding analyses and insights, and a concluding discussion outlining future work.

Description of assessment methods: used to evaluate the effectiveness of the contribution
To assess the framework, we evaluated the number and depth of the technologies and methods integrated across the end-to-end pipeline: from human muscle signal acquisition to robotic control and movement. The overall cost of the framework was also considered to measure its accessibility and practicality.
In terms of STEM education outcomes, particularly in computational and engineering thinking, we assessed the quality of the collected data, the soundness of feature selection, and the accuracy of the developed classifiers to measure students’ conceptual understanding. Also, the accuracy and speed of robot movements in the test environment were examined to evaluate students’ proficiency in hardware–software integration.

A statement of results:
Robotics, as an inherently interdisciplinary field, can play a vital role in fostering early interest in science, technology, engineering, and mathematics (STEM) among high school and first-year college students. However, many students are exposed to robotics primarily through theoretical instruction with little real-world experience. Considering their limited exposure to engineering and computer science, a well-designed robotics platform, supported by structured, hands-on projects, can bridge the gap between abstract concepts and practical applications.
To this end, our team, comprising a faculty mentor, one high school student, and one first-year undergraduate, developed an experiential robotics learning framework during the summer of 2025. The framework integrates electromyography (EMG) sensing, TurtleBot3 mobile robotics, the Arduino IDE, and a personal computer for AI-related computation. Using MyoWare 2.0 EMG sensors placed on the hand, forearm, and shoulder, muscle activation signals were measured and transmitted to an Arduino microcontroller, which converted the analog signals to digital form for further analysis. Students extracted statistical and frequency-domain features, and applied AI/machine learning (ML) algorithms in Google Colab to classify EMG data. Once a robust classifier was trained, it processed live EMG data, generated predictions, and transmitted corresponding control commands to the ROS 2 environment, enabling real-time control of the TurtleBot3. This process completed an end-to-end pipeline from human muscle signals to robotic control and movement.

Using this framework, students successfully developed two high-performing classifiers for EMG data from different muscle groups. The shoulder-based motion classifier achieved approximately 90% accuracy, while the forearm-based motion classifier achieved about 81%. Together with managing communications and control devices, they were able to achieve live, multidirectional control of the TurtleBot3 based on real-time EMG input (a demo is available at https://www.youtube.com/watch?v=qUq31F_yBtc). These results validate both the technical robustness of the framework and the effectiveness of its instructional design, demonstrating the feasibility of transforming human physiological data into robot actions through engineering, data science and AI/ML.

The presented research addresses “undergraduate research and the integration of research and education.”