Disturbance rejection for U.A.S. aircraft using bio-inspired strain sensing

Abstract

A bio inspired gust rejection mechanism based on structural inputs is proposed. Insect wings possess a wealth of sensor systems which typically consist of fast reflexive neuronal paths. Stretch and strain sensors on insect wings are used for flight control and can be found across many species. These are used for monitoring of bending and torsion during flight. The fast reflexive and proprioceptive mechanisms based on strain sensing found in nature are the inspiration for this work. A strain feedback controller allows for anticipation of the onset of rigid body dynamics due to gust perturbations. This anticipation stems from sensing of higher order states and the possibility of reacting before lower order states are reached. High bandwidth inner loop compensation is therefore enabled. Forces and moments are proportional to wing strain patterns and can be used in fast reaction inner loops. Strain sensors are used for providing an indirect estimation of the differential forces applied to the aircraft wing and therefore to the aircraft rigid body. These sensors can be distributed over the surface of the aircraft wing to encode multiple degree of freedom disturbances. Sensor locations for disturbance rejection are determined based on metrics associated to the observability Grammian. The locations are preselected based on modal energy analyses and are chosen according to wide field integration patterns. A model for wide field integrated strain based on mass participation factors is proposed as well as one which is based on the physics of the forces and moments acting on the wing producing strain patterns which can be used for disturbance rejection. Models of the differential forces via strains on the wings are proposed. Strain feedback was implemented in four platforms under different types of disturbances. The platforms consisted of a glider, a quadrotor, a wing section for wind tunnel testing and an RC airplane with a full span wing. The disturbances included discrete gusts as well as turbulence. The results of using strain feedback showed not only to be faster than IMU estimations but also to be better when compared to a classical attitude controller implementation.