Incremental Nonlinear Control of Quadrimotors: Sensitivity Analysis and Experimental Validation in Attitude and Trajectory Tracking
Name: FABRÍCIO CUSTODIO JACINTO
Publication date: 30/03/2026
Examining board:
Name |
Role |
|---|---|
| JOSE LEANDRO FELIX SALLES | Examinador Interno |
| JOSÉ REGINALDO HUGHES CARVALHO | Examinador Externo |
| RAFAEL DE ANGELIS CORDEIRO | Presidente |
Summary: This dissertation presents the development and evaluation of incremental nonlinear control strategies for
attitude and trajectory tracking of micro-quadrotors. Due to the low inertial parameters of these platforms,
classical controllers often face stabilization challenges under external disturbances or during aggressive
maneuvers. To overcome these limitations, this work applies and compares Cascaded Incremental Nonlinear
Dynamic Inversion (C-INDI), Incremental Backstepping (IBKS), and both classical and incremental Nonlinear
Dynamic Inversion (NDI/INDI). These approaches aim to bypass the need for exact mathematical models by
robustly handling the system’s dynamic uncertainties.
Practical experiments were conducted and divided into two main stages. Initially, focusing on attitude control,
the C-INDI and IBKS controllers were embedded into the Bitcraze Crazyflie 2.1 micro-quadrotor. The flight
test scenarios involved subjecting the aircraft to severe wind disturbances and executing aggressive maneuvers.
Subsequently, the NDI and INDI strategies were implemented on the Parrot Bebop 2 platform for trajectory
tracking, where a parametric sensitivity analysis was also conducted to evaluate the impact of uncertainties
on the performance of the strategies.
The results demonstrate that the proposed techniques allow the drones to sustain efficient flight under adverse
conditions. In the attitude tests, both incremental strategies adapted quickly to continuous disturbances,
with the IBKS controller achieving a slightly faster dynamic response. In the trajectory tracking stage, the
sensitivity analysis showed that the INDI strategy is significantly more robust to modeling uncertainties
compared to its non-incremental version. However, it was observed that the performance of incremental
techniques is highly dependent on sensor quality and sensitive to measurement noise. This characteristic
highlights the need for improvements in filtering and state estimation algorithms, indicating directions for
future work focused on enhancing the reliability and safety of autonomous navigation.
