Towards Plug-and-Fly Quadrotors: Minimal Identification and Automatic Tuning with Incremental-Based Flight Control

Uncrewed Aerial Vehicles (UAVs) commonly rely on accurate dynamic models and extensive system identification to achieve high-performance flight control, which often requires expert knowledge and complex experimental trials, representing significant barriers to the rapid deployment of new platforms. Furthermore, these processes must be repeated whenever the configuration changes due to modifications in geometry, payload, or propulsion systems, limiting the adaptability and ease of reconfiguration of autonomous aircraft. Incremental control strategies, such as Incremental Nonlinear Dynamics Inversion (INDI), overcome much of this complexity by requiring only the actuator control effectiveness rather than a complete dynamics model. Accounting for such advantage, this paper proposes a dedicated framework for quadrotors controlled with incremental strategies that combines minimal system identification with automatic controller tuning, enabling rapid controller deployment with limited prior knowledge of the aircraft.

The underlying methodology relies on two sequential steps: identification of the actuator effectiveness, followed by automatic tuning of the INDI controller. The identification process estimates the control effectiveness matrix through a dedicated procedure, in which each attitude control channel is independently excited, and the resulting estimated angular accelerations are processed through linear regression to estimate the different components of this matrix. Subsequently, the gains of the linear feedback component of the controller are automatically tuned by minimizing a multi-objective cost function that accounts for transient response, oscillatory behaviour, and control effort.

The proposed framework is first validated in simulation, demonstrating accurate estimation of the control effectiveness and successful controller adjustment. Experimental validation is then performed in two stages: identification and tuning on a dedicated gimbal test bench, followed by free-flight experiments. The obtained results demonstrate the feasibility of an almost fully autonomous configuration of the INDI-based flight controller with minimal modelling and identification effort, achieving a stable autonomous flight without any prior knowledge of the characteristics of the UAV.