Pathways to Learning: Exploring High School Students' Learning of AI-Powered Educational Robotics

Synthesis: A qualitative study of 15 high school students in a semester-long, project-based robotics + AI curriculum (mBot kits + image recognition ML) reveals that students develop understanding of Educational Robotics and AI through three epistemically plural pathways — engaging in real-world robotics practices (troubleshooting hardware, sensors, and physical conditions), designing (building and iterating costumes for their robots), and playful/creative expression (personalizing code and exploring AI features). Grounded in Constructivist theory and epistemological pluralism, the study argues these diverse forms of engagement are legitimate ways of knowing in computing — not mere stepping stones toward "real" coding — and offers alternative entry points for broadening participation in K 12 CS Education.

Core Finding

Drawing on constructionist theory and epistemological pluralism (Turkle & Papert, 1992), this study reframes learning in Educational Robotics and AI as an epistemically plural endeavor: high school students developed meaningful understandings of robotics and AI not through a single, code-centric route, but through three distinct and legitimate pathways — engagement in real-world robotics practices, designing, and playful/creative expression. The qualitative analysis (interviews, 31 hours of field notes, student artifacts, final presentations) shows students learned robotics (motion planning, programming, robot structure, sensors) and AI (machine learning, sensor-based perception) through these diverse forms of engagement, demonstrating that pluralistic approaches do not come at the expense of disciplinary goals.

The study positions these three pathways not as preparatory steps toward "real" computing, but as alternative endpoints in computing education — arguing for expanding what counts as valued CS knowledge and offering meaningful entry points for learners who may not connect with traditional, lecture-based computer science pedagogies.

Learning Through Engaging in Real-World Robotics Practices

Working with physical robots required coordinating computational logic with hardware, operational factors, and physical conditions. Students encountered and managed "material variability" and system "pushback" — wheel calibration, battery level, initial positioning, lighting, and surface reflectivity — which became sites of learning rather than distractions.

Hardware learning: Students diagnosed why robots veered off course or failed to turn exactly 90 degrees (e.g., Robert attributed drift to one wheel being more sensitive; Luna suspected battery level).
Operational learning: Students learned that initial placement shapes trajectory — Luna strategically positioned her robot to compensate for predictable drift.
AI learning through physical setup: James realized his color sensor failed to detect blue ("maybe blue was too dark") and reasoned about how lighting and surface reflectivity affect robot perception; Eric learned that background clutter and overhead lighting degrade the data quality of his image-recognition ML model.

This theme reflects learning through empirical engagement with material systems — a concrete, situated way of knowing grounded in the "dance of agency" between learners and the physical world.

Learning Through Designing

Students designed, built, and iteratively refined costumes for their robots (e.g., a Star character from Disney's Wish, a SpongeBob figure). The design process became a way of reasoning about robot structure and motion planning.

Students learned to treat form, weight, placement, and mobility as mutually constitutive within a functioning system — not as mere decoration.
E.g., Mike's SpongeBob costume sat on the wheels and was too heavy; he iteratively cut material around the wheels and swapped tape/glue for lighter hot glue; Luna connected costume weight to the robot's inability to turn exactly 90 degrees.
This aligns with learning-through-design frameworks (Kafai & Resnick, 2012), where trade-offs about form and structure are central to understanding how robots function.

Learning Through Playful and Creative Expressions

Students personalized code and explored AI/robot features beyond assigned tasks — a distinct, curiosity-driven way of constructing knowledge.

Luna programmed the display to print "I love my best friend" and implemented rainbow LED effects; Kim recorded her favorite song for the robot to play back.
Kim tweaked the ultrasonic sensor code so the robot turned left when sensing an object, then played by placing her foot in front of it — leading to the insight that "the robot didn't have eyes… the robot can only sense things," a foundational AI understanding of sensor-driven perception.
Jacob trained the image-recognition model to recognize him and coded the robot to say "It's J!"

This theme highlights exploratory and expressive engagement as a legitimate epistemic route in computing, in contrast to problem-solving or constraint-satisfaction learning.

Relevance to the Wiki

This paper is a significant contribution to the Educational Robotics and Project Based Learning concepts. It provides one of the clearest empirical illustrations of epistemological pluralism in action: three distinct, legitimate ways that high school learners develop understanding of robotics and AI, grounded in Constructivist theory. For the wiki's focus on AI Education and K 12, it demonstrates a concrete, qualitative account of how students develop critical AI understanding through tangible ML experiences (training image-recognition models, reasoning about data quality and sensor limitations) — a valuable complement to the field's emphasis on attitudes and skills. The study also speaks to broadening participation in CS Education by legitimizing design, play, and real-world practice as alternative endpoints, relevant to Student Engagement, Motivation, and Creativity.

Notably, the study is qualitative — it deliberately does not report quantified learning gains, instead documenting how learning unfolds through diverse forms of engagement. This should not be read as evidence against learning gains from Educational Robotics, but as an account of the pathways through which such learning occurs.

Connected Concepts

Connected Articles

Aaiwa AI Authentic Assessment Metacognition 2026 — AI and Authentic Assessment
Aaai2026 Prompting Literacy K12 — Learning to Use AI for Learning: AI Literacy Module for K-12
AI Education Global Capacity — What AI in Education Needs Next: Lessons from Youth Leaders

Citation

McLaughlin, G., Novak, E., Ahmadi, S., Li, J., Guo, Y., Liu, R., & Borgerding, L. (2026). Pathways to Learning: Exploring High School Students' Learning of AI-Powered Educational Robotics. Educational Technology Research and Development.

View Source View Changes Edit