In this paper, reinforcement learning is implemented to train a neural network-based feedback controller for turbulent gust rejection on a flexible wing. The aeroelastic model of the wing is modeled by coupling the unsteady vortex lattice method for unsteady aerodynamics with finite-element based structural dynamics. Thereafter, reinforcement learning via the proximal policy optimization algorithm is used to train a neural network to minimize gust-induced tip deflections by directly manipulating the control surface. Results from simulation show that the trained policies are able to reduce the mean-squared-error in the tip displacements compared to the open-loop responses. Additionally, observing the incoming gust magnitudes, measured by an alpha probe, significantly improves the performance of the controller by allowing it to take preemptive actions. The reinforcement learning policy is able to better adapt to and learn complex dynamics, resulting in better overall performance when compared to PD and LQG controllers.

This study investigates a system of hexacopter type drones with evolvable morphologies and learnable controllers. The combination of morphological evolution and reinforcement learning is shown to produce unconventional drones that significantly outperform the traditional hexacopter on several tasks that are more complex than previously considered in the literature. In addition, novel metrics are introduced and new analyses are conducted on the interaction between morphological evolution and learning, uncovering previously unidentified effects.

In this paper, the stability margins of a flying wing unmanned aerial vehicle (UAV) undergoing body-freedom flutter (BFF) are investigated. The stability margins are computed using a coupled aeroelastic model comprising finite-element-based structural dynamics with the unsteady vortex lattice method. Two primary modes of BFF are observed, a plunge-dominant mode and a pitch-dominant mode, which are closely related to the structural stiffness and inertial properties. In essence, increasing bending or torsional stiffness increases the tendency of plunge or pitch-dominant mode, respectively. In addition, flutter frequencies of the pitch-dominant mode are higher than the plunge-dominant mode, and the flutter speed is found to be highest near the point of transition between the flutter modes. It is suggested for structural properties of flying wing UAVs to be designed such that the flutter mode is near the transition point for increased stability margins. Additionally, plunge-dominant flutter mode is preferred due to its lower frequency, which will be easier to mitigate. The findings from this study provide insights into the design of flying wings on how stiffness/ mass distribution affects the coupling between elastic and rigid-body modes leading to BFF.