Abstract:The strong nonlinear dynamics of tail-sitter vertical takeoff and landing (VTOL) unmanned aerial vehicles (UAVs) have consistently constrained their applications. Particularly, the widely-varying flight speed and attitude during the transition flights pose substantial challenges for the controller design. In order to gain a clear understanding of the nonlinear transition flights, the long- and short- period dynamic characteristics of tail-sitter UAVs are investigated for the first time. The results reveal that the short-period mode exhibits either sluggish response or excessive overshoot in different transition stages, primarily due to the widely-varying damping ratio and natural angular frequency. Even worse, the long-period mode can diverge due to positive poles. To address these issues, a systematic controller is proposed. First, the transition trajectory is optimized and the optimal control inputs are employed in a feedforward manner to trim the nominal forces and moments. On this basis, a discrete-time linear quadratic regulation (LQR) controller with a predefined decay rate is developed to position the desired poles, and a novel angular acceleration estimation method is introduced to compensate for unmodeled dynamics. Simulations under different uncertainties indicate that the proposed control method performs better in both transition trajectory tracking and uncertainty suppression, compared to the PID and incremental nonlinear dynamic inversion (INDI) controllers.
What problem does this paper attempt to address?
### Problems the paper attempts to solve
The paper aims to solve the strong non - linear dynamic problems encountered by tail - sitter vertical take - off and landing (VTOL) unmanned aerial vehicles (UAVs) during flight mode transitions. Specifically, during the transition from hovering to cruising (forward transition) or from cruising to hovering (reverse transition), the flight speed and attitude of tail - sitter VTOL UAVs will change significantly, which poses a great challenge to the controller design. These problems are mainly reflected in the following aspects:
1. **Response problems in short - period mode**: The short - period mode may exhibit sluggish response or excessive overshoot in different transition stages, mainly due to large changes in damping ratio and natural angular frequency.
2. **Stability problems in long - period mode**: There may be positive poles in the long - period mode, causing the system state to diverge and thus affecting flight stability.
To meet these challenges, the paper proposes a systematic control method, which consists of the following three parts:
1. **Trajectory optimization and feed - forward control**: Optimize the transition trajectory to determine the optimal control input for adjusting the nominal forces and moments.
2. **Discrete - time linear quadratic regulator (LQR) control**: Develop a discrete - time LQR controller with a predefined decay rate for locating the desired poles.
3. **Robust compensation based on angular acceleration estimation**: Introduce a new angular acceleration estimation method to compensate for unmodeled dynamics and external disturbances.
### Main contributions
1. **Dynamic characteristic analysis**: Through the linearization analysis of the transition dynamics, the long - period and short - period dynamic characteristics of tail - sitter VTOL UAVs are studied for the first time. The results show that the damping ratio and natural angular frequency of the short - period mode change greatly in different transition stages, resulting in sluggish response or excessive overshoot. Meanwhile, the positive poles in the long - period mode may cause the system state to diverge.
2. **New discrete - time LQR controller**: Develop a discrete - time LQR controller with a predefined decay rate to ensure the minimum decay rate and the desired pole configuration during the transition process, and give an iterative algorithm for solving the discrete - time Riccati equation.
3. **Angular acceleration estimation method**: Propose an angular acceleration estimation method based on a one - step lag nominal model, which can effectively estimate and compensate for unmodeled dynamics and external disturbances.
### Simulation verification
System simulation verifies the effectiveness of the proposed control method. The simulation results show that the dynamic adjustment error of the proposed angular acceleration estimation method is only 21.7% of that of the Kalman filter based on the constant acceleration model, and it can estimate the actual angular acceleration timely and accurately. The controller evaluation simulation shows that the proposed control method exhibits superior performance under external and internal uncertainties. Compared with the PID controller, the proposed control method reduces the average relative error in attitude tracking by at least 60%. Compared with the INDI controller, the proposed control method can achieve better attitude tracking while requiring less control power, and the average actuator saturation time is reduced by at least 34%.
### Conclusion
Through dynamic characteristic analysis and the design of a systematic control method, the paper effectively solves the stability and robustness problems of tail - sitter VTOL UAVs during the transition process, providing theoretical and technical support for practical applications.