Robust Adaptive Control Algorithms for Vertical Take-Off and Landing Autonomous Unmanned Aerial Vehicles
Abstract
Goals of the proposed research: This research is aimed at the design of robust, adaptive, scalable control systems for long-range vertical take-off and landing (VTOL) unmanned aerial vehicles (UAVs) to be employed by the US Navy in operations that range from surveillance to payload delivery. The proposed control algorithms will exploit both the UAV’s propellers and its wings to take off, land, and hover vertically like multi-rotor UAVs, and fly horizontally like conventional fixed-wing aircraft. These control systems will guarantee robustness to uncertainties in the payload’s inertial properties, such as its contribution to the displacement of the UAV’s center of mass and moment of inertia, coarse modeling of the UAV’s aerodynamic forces and moments, and faults and failures in the propulsion system due to wear and tears or damages occurred during the mission. A careful dynamics-based, control-oriented dynamical modeling of the UAV’s dynamics, which is agnostic of the UAV’s actual dimensions, will allow the proposed control systems to be operative on aircraft of different scales, ranging from Class 1 to Class 5. Therefore, although the proposed control algorithms will be tested on Class 1 tail-sitter quad-biplanes and fixed-wing tilt-rotor UAVs only, the control algorithms produced as part of this research will also apply to other vehicles such as the MUX UAV capability or the V-22 Osprey. Lastly, the proposed controls will augment existing control algorithms for VTOL aircraft and will increase their robustness margins. Summary of the state of the art and technical challenges: Long-range VTOL UAVs, that is, multi-rotor UAVs equipped with fixed wings, are relatively new and numerous challenges need to be addressed to enable their systematic use in Navy operations. Indeed, the great majority of existing autopilots do not exploit these vehicles’ nonlinear dynamics and hence, poorly handle the transition between vertical and horizontal flight regimes. Additional complexity is given by the fact that there always exists at least one configuration for which these vehicles are uncontrollable. Uncertainties in the aerodynamic model and lack of accurate information on the UAV’s dynamics due to unmodeled payloads further challenge the ability of existing controllers for long-range VTOL UAVs to guarantee satisfactory flight performance. Lastly, the problem of creating control algorithms that guarantee robustness to faults and failures and are sufficiently fast to be implemented on state-of-the-art single board computers provides an additional challenge. Novelty and uniqueness of the proposed research: The use of a learning technology, whereby the controller adapts its performance to the vehicle’s actual response and considers user-defined analytical models as initial conditions for the learning mechanism, allows exploiting the UAV’s nonlinearities to improve its trajectory tracking performance. Additionally, considering all dimensional properties of the UAV, such as the location of its center of mass and its aerodynamic center, as unknowns is a key factor to guarantee scalability of the proposed control algorithms. The use of the dead-zone function in the proposed control laws guarantees controllability of the UAV in any configuration. Past experiences of the principal investigator with similar algorithms shows their efficiency and the use of this technology on existing single-board computers for UAVs. Impact of the proposed research: The proposed research will enable the use of long-range VTOL UAVs to be employed by the US Navy for surveillance and payload delivery missions. Furthermore, this research will produce control algorithms that can be implemented also on aircraft of great relevance to the US Navy such as the MUX UAV and the V-22 Osprey.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 28, 2021
- Source ID
- N004212110004
Entities
People
- Andrea L Afflitto
Organizations
- United States Navy
- Virginia Tech