The Team America Rocketry Challenge (TARC) requires student teams to design rockets capable of meeting strict altitude and flight-time requirements. Traditional rocket prediction methods rely heavily on physics-based simulations and repeated experimental launches, which may not fully account for real-world aerodynamic variations. This study investigates the application of machine learning as a supplementary tool for predicting rocket flight performance in model rocketry. A physics-based flight simulator was first developed using classical mechanics, thrust modeling, and aerodynamic drag equations to estimate rocket trajectory, apogee, and flight time. A Random Forest Regression model was then trained using approximately 500 flight samples containing rocket parameters and measured apogees. The machine learning model used rocket weight, rocket height, launch angle, and wind speed as input variables. Model performance was evaluated using Mean Absolute Error (MAE). Results showed that the physics-based simulation predicted an apogee of approximately 697.77 ft, while the machine learning model achieved an MAE of approximately 21.18 ft and predicted an apogee of 759.52 ft for the tested rocket configuration. The relatively small difference between the two prediction approaches demonstrates that machine learning can effectively approximate complex flight behavior while significantly reducing computation time. The findings suggest that machine learning can serve as a practical supplementary tool alongside traditional physics-based simulations for improving rocket flight prediction and design optimization in TARC.

Xinyu Du. The Application of Machine Learning-Driven AI in Model Rocketry: Improving Flight Simulation and Data Analysis in TARC. Academic Journal of Computing & Information Science (2026), Vol. 9, Issue 6: 1-8. https://doi.org/10.25236/AJCIS.2026.090601.

[1] Aerospace Industries Association. “Competition Guidelines.” Team America Rocketry Challenge, Available from: https://www.rocketcontest.org

[2] IBM. “Machine Learning.” IBM Technology Topics, Available from: https://www.ibm.com/topics/ machine-learning.

[3] Fears, Darrell. “Scientific Machine Learning Paves Way for Rapid Rocket Engine Design.” UT Austin News, 16 Apr. 2020, Available from: https://news.utexas.edu/2020/04/16/scientific-machine-learning-paves-way-for-rapid-rocket-engine-design.

[4] Kamath, Sreejith, et al. “Amortized Inference for Rocket Flight Performance.” arXiv, 2025, Available from: https://arxiv.org/html/2512.22248v1.

[5] Pedregosa, Fabian, et al. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research, 2011. Available from: https://scikit-learn.sourceforge.net/dev/about.html

[6] McKee, Martin Jay. “Newsletter #616.” Apogee Rockets Education, Available from: https://apogeerockets.com/education/downloads/Newsletter616.pdf.

[7] Harris, Charles R., et al. “Array Programming with NumPy.” Nature, 2020, 585(7825): 357–362.

[8] Hunter, John D. “Matplotlib: A 2D Graphics Environment.” Computing in Science & Engineering, 2007, 9(3): 90–95.

[9] Virtanen, Pauli, et al. “SciPy 1.0: Fundamental Algorithms for Scientific Computing in Python.” Nature Methods, 2020, 17(3): 261–272.