Identification of quadcopter hovering using experimental data

Obtaining the plants' mathematical models using physical laws is essential in the development and design of suitable controllers. However, sometimes these models can be complex, inaccurate, or difficult to find. In such cases, mathematical models can be build using experimental data that approximate the real system's behavior. The quadcopter is one such example where the relationship between its inputs and outputs is nonlinear, complex, and dependent on several difficult-to-measure parameters. In this paper, a quadcopter is assembled, tested, and used in several flight experiments where the Input-Output (I/O) data is collected and used to develop accurate mathematical models of the quadcopter for hovering operation. Auto-Regressive with eXogenous (ARX) based models, for both Single-Input Single-Output (SISO) and Multi-Input Multi-Output (MIMO), are developed and analyzed. Results show that the developed models provide good accuracy when compared to the experimental measurements.