Constrained estimation-based nonlinear model predictive control for UAV-elastic tool interaction

Interaction between an unmanned aerial vehicle (UAV) and an object or environment is a promising field for real-time applications. However, this state of the art brings challenging control problems. In order to handle the problem called stiff equation as well as bouncing, a tailored elastic tool based on a passive spring is proposed which can provide a safe interaction with the environment. Due to having constraints, e.g., the touching sensor's upper force bound and limited space for the operation environment, a constrained optimization-based algorithm is considered. The full system state, including an unmeasured vertical force, is estimated by nonlinear moving horizon estimation (NMHE) after each new measurement becomes available. The estimated values are then fed into the nonlinear model predictive control (NMPC) which provides the total force and three angular positions. Thanks to the vertical force information given by the NMHE, the presented interaction controller is able to interact with the ceiling in a predefined range. As opposed to multi-model approaches, a modeling, estimation, and control of the whole system are presented with a centralized approach. However, the loop consisting of an NMPC and NMHE is closed in an average of 5 ms.