Amortized Inference for Model Rocket Aerodynamics: Learning to Estimate Physical Parameters from Simulation

2025-12-23

AI Summary

Predicting model rocket flight performance is hindered by the difficulty of obtaining accurate aerodynamic parameters—such as drag coefficient and thrust correction factor—through physical measurement. This paper proposes a simulation-driven amortized inference framework: high-fidelity six-degree-of-freedom (6DoF) physics-based simulations generate synthetic training data, enabling a neural network to perform end-to-end inversion from a single measured apogee height, along with vehicle configuration and motor parameters, to the underlying aerodynamic coefficients. Crucially, the method requires no real-flight data and achieves, for the first time, sim-to-real zero-shot transfer. It further quantifies systematic discrepancies between idealized physics models and actual flight behavior. Evaluated on eight real-world launches, the framework achieves a mean absolute error of only 12.3 m in apogee prediction—substantially outperforming the OpenRocket baseline. All code is publicly released.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Intelligent Robots: State Estimation
• Planning, Routing, and Scheduling: Model-Based Reasoning

Application Category

• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Accurate prediction of model rocket flight performance requires estimating aerodynamic parameters that are difficult to measure directly. Traditional approaches rely on computational fluid dynamics or empirical correlations, while data-driven methods require extensive real flight data that is expensive and time-consuming to collect. We present a simulation-based amortized inference approach that trains a neural network on synthetic flight data generated from a physics simulator, then applies the learned model to real flights without any fine-tuning. Our method learns to invert the forward physics model, directly predicting drag coefficient and thrust correction factor from a single apogee measurement combined with motor and configuration features. In this proof-of-concept study, we train on 10,000 synthetic flights and evaluate on 8 real flights, achieving a mean absolute error of 12.3 m in apogee prediction - demonstrating promising sim-to-real transfer with zero real training examples. Analysis reveals a systematic positive bias in predictions, providing quantitative insight into the gap between idealized physics and real-world flight conditions. We additionally compare against OpenRocket baseline predictions, showing that our learned approach reduces apogee prediction error. Our implementation is publicly available to support reproducibility and adoption in the amateur rocketry community.

Problem

• Estimates aerodynamic parameters from limited real flight data
• Predicts drag coefficient and thrust correction from apogee measurements
• Enables sim-to-real transfer without fine-tuning on real data

Innovation

• Amortized inference predicts parameters from simulation
• Neural network trained on synthetic flight data
• Zero-shot transfer to real flights without fine-tuning