Data-Driven Dynamics Prediction of a Quadcopter Using Koopman Operator

This paper introduces a novel prediction method for a quadcopter model based on Koopman operator theory. Due to the inherently high non-linearity of the quadcopter dynamics, the proposed method leverages the Koopman operator to transform the nonlinear system into a higher-dimensional space, thereby constructing a linear surrogate model. This linear surrogate facilitates the application of linear prediction techniques to the original nonlinear system. The approach utilizes the Koopman-derived model to accurately predict system states. Comprehensive simulations of the quadcopter model are conducted to validate the efficacy of the proposed method. The results demonstrate both the predictive accuracy and the robust performance of the method in handling the complex dynamics of the quadcopter.