AR-Enhanced Physics Lab
Elizabeth Childs, Kongmeng Her, Kristen Pilner Blair, Allison M. Okamura, James A. Landay
- Role
- Lead author. Designed and built the AR lab and led the 52-student study, and mentored undergraduate researcher Kongmeng Her.
- Study
- Between-subjects study with 52 students comparing the AR lab to a traditional lab
- Tools & techniques
- Augmented reality, computer vision, real-time data visualization, laser cutting, between-subjects user study, and Bayesian analysis.
Hands-on physics labs are supposed to teach how science works, but students often spend the lab hand-counting swings and reading angles instead of thinking about their data. This project keeps the real pendulum and adds AR measurement tools on top of it: a protractor anchored at the pivot, a live graph of energy moving between potential, kinetic, and dissipated forms, and an automatic period counter. In a study with 52 students, the AR lab improved scientific reasoning over the traditional lab (Cohen’s d = 0.63, Bayes factor 17.44) while maintaining interest and content mastery.
Concept and Design
The AR system was developed as a mobile application layered over a real physical pendulum. It was designed to act as a cognitive aid by embedding measurement tools and visualizations directly in the experimental space:
- AR Protractor – Anchored at the pivot point, a semi-transparent digital protractor displayed the pendulum’s angle in real time, along with the instrument’s uncertainty.
- Energy Graph – A dynamic visualization showing the flow of potential, kinetic, and dissipated energy during motion, making normally abstract energy concepts visible.
- Period Counter – Automated timing of oscillations, with both total counts and average periods, relieving participants from hand-counting and reducing measurement error.
- Lab Notebook Interface – A paired digital notebook that stored results, computed statistics (including t-tests), and displayed uncertainty values alongside measured data.
By combining real instruments with AR overlays, the system was designed to reduce extraneous cognitive load, support distributed cognition across digital and physical artifacts, and provide meta-representations that connected abstract physics concepts with real-time measurements.
Results
The results showed that AR-enhanced labs led to significantly greater improvements in scientific reasoning than traditional lab setups. Bayesian analysis produced a Bayes Factor of 17.44 favoring the AR condition, with a medium-to-large effect size (Cohen’s d = 0.63), indicating that participants using AR gained more from the experience. Students in the AR group engaged more deeply with their data: they were more likely to notice differences in energy dissipation, to reference uncertainty in their reasoning, and to trust their results. Energy transfer emerged as a particularly salient concept, with AR participants twice as likely to mention it as an insight compared to the control group. While both groups grasped the main physics concept—that larger angles affect pendulum period—the AR group showed stronger reasoning about why this was the case.
The findings suggest AR can act as a cognitive partner in laboratory learning, shifting student attention from procedural demands to critical evaluation of evidence.