Alocação de controle desacoplada rápida em sistemas de controle superatuados

Abstract

Over-actuated systems usually require nonlinear control allocation methods to map Virtual Control Actions (VCAs) into Real Control Actions (RCAs). This process requires computational efforts sometimes not available on embedded robotic platforms. It is in this context that this work presents the design of an Quadrotor Tilt-Rotor (QTR) through a new concept of control allocation with uncoupled RCAs, where the initial nonlinear system is divided into partially dependent linear subsystems with fast and robust convergence. For this purpose, the RCAs are divided into smaller sets, used sequentially to linearize and solve the system. The reduction of the initial nonlinear control effectiveness matrix is improved by selecting each subset in a different arrangement of VCAs. However, the choice of this arrangement may lead to absence, partial or full superposition of VCAs in the subsystems. The technique was validated through mathematical tutorial cases, QTR simulation tests and open field flight and gyroscopic test bench experimental tests. Finally, the control allocation technique proved to be reliable, robust, efficient and applicable in the QTR when there is full superposition of VCAs between the subsystems.