Jason L. Forshaw

Indoor Experimentation and Flight Test Results for a Twin Rotor Tailsitter Unmanned Air Vehicle

This paper continues the progressive research on the QinetiQ Eye-OnTM, a twin helicopter rotor, twin elevon tailsitter, which remains unique from other tailsitters in past literature, and provides a background to initial experimental work with the vehicle. As part of a staged flight test campaign, the vehicle is initially flown manually then progressive indoor testing and outdoor testing is performed to develop the technology necessary for autonomous flight; this paper addresses the first two endeavors. Manual flight testing confirms transitioning between vertical and horizontal modes of flight, although difficult, is feasible. The indoor experimentation is performed on a motion capture system; its architecture is firstly explored then both attitude and thrust testing is performed to validate and tune the control laws. Thrust testbed results explore fully the nature of twin rotor systems in terms of collective, RPM, swashplate motion and control signals. From observing that thrust progresses through four distinct stages as collective increases, thrust management schemes are developed for VTOL, vertical, and horizontal flight segments. The results are a significant step in the development of a fully autonomous control architecture for this unique tailsitter in outdoor flight.

Transitional Control Architecture, Methodology, and Robustness for a Twin Helicopter Rotor Tailsitter

This paper continues the progressive research on the QinetiQ Eye-OnTM and considers transitioning architecture, methodology and robustness. The UAV is ‘over-defined’ in terms of its control inputs due to having both elevons and twin helicopter assemblies. This research presents a novel control and navigation architecture for the vehicle (that coordinates both horizontal and vertical modes of flight, navigation and transitioning) capable of handling the overdefined set of control inputs. The research also presents a new methodology for transitioning in a closed-loop manner, in comparison to many other research groups that use open-loop transitioning. Both horizontal to vertical and vertical to horizontal transitions are examined in simulation with specific consideration for the time to transition and horizontal flight speed. Additionally, analysis of transitioning robustness is undertaken by considering sensor inaccuracies and gusts in a stochastic simulation. Finally, a motion-capture testbed setup is presented for initial indoor testing and a tandem helicopter rotor platform is constructed for use in vertical flight testing for the tailsitter. Results demonstrate that transitions on this unique class of tailsitter are accurate and robust to real-life external disturbances.

Online Evolutionary Swarm Algorithm for Self-Tuning Unmanned Flight Control Laws

ap, bp, cp = fitness function weightings B = body frame Bcm = center of mass of the vehicle in B D = D gain for proportional integral derivative e = error, generic F e = fitness function I = inertial frame IBcm B = inertial tensor ofBcm with respect toB, kg · m, expressed in B I, ki = I gain for proportional integral derivative P, kp = P gain for proportional integral derivative qe = quaternion error from B to I r = position, generic, m u = input to the plant v = velocity, generic, m∕s v, v, v = velocity of Bcm with respect to I, m∕s, expressed in B x, y, z = position of Bcm from Io, m, expressed in I xc, yc, zc = position setpoints, m δc, δr, δp, δy = thrust, roll, pitch, yaw commands ζ = damping ratio φ, θ, ψ = Euler angles from B to I, deg ψc = yaw setpoint, deg ω = linear control bandwidth ω B = angular rate from B to I, rad∕s, expressed in B

OQTAL: Optimal quaternion tracking using attitude error linearization

The use of quaternions or quaternion error attitude control strategies for unmanned aerial vehicles (UAVs) is commonplace. Quaternion tracking error control is usually presented in rather theoretical works, where the control algorithm is almost exclusively chosen to be a suboptimal one. However, the application of optimal control techniques is usually associated to simplified attitude models frequently aimed at solving real-life problems. The work presented in this paper aims to formally merge the development of a complete theoretical quaternion error model with an optimal control strategy. Moreover, the application of optimal control algorithms to a fully defined quaternion error state-space model and the validation of the same in a real-time experimental setup is the focus of this research. The result is a novel controller named Optimal Quaternion Tracking of Attitude Error Linearization (OQTAL). The paper provides a comprehensive proof of stability, full simulation validation for a planetary landing gravity turn trajectory, and evidence of repeatable experimental work for a real-time quadrotor UAV on a motion capture testbed. OQTAL is compared with proven optimal forms of (PID) proportional-integral-derivative control and linear quadratic regulator control and is shown to have a 10%-20% reduction in error for the experimental setup trajectory tracking trials and an even larger tracking error reduction in the gravity turn simulation trials. Furthermore, for close tracking conditions, OQTAL behaves almost like a linear and time-invariant system, therefore requiring limited computation time for performing the trajectory tracking.

Transitional Control Architecture and Methodology for a Twin Rotor Tailsitter

B = body frame Bcm = center of mass of the vehicle in B e = error, generic Fai = aerodynamic forces for axis i, expressed in B, N Fgi = gravity forces for axis i, expressed in B, N Fpi = propulsive forces for axis i, expressed in B, N h = altitude, m I = inertial frame Mai = aerodynamic moments about axis i at Bcm, expressed in B, N · m Mpi = propulsive moments about axis i at Bcm, expressed in B, N · m p, q, r = angular rate, deg ∕s tp = transitioning period, time taken to transition, s u, v, w = velocity of Bcm with respect to I, expressed in B, m∕s uc = velocity setpoint command, m∕s vc0 = horizontal cruise speed, m∕s α, β = angle of attack, sideslip angle, deg δc, δw = thrust, collective, rpm commands δe, δa = elevator (via elevon), aileron (via elevon) deflection commands δr, δp, δy = roll, pitch, yaw commands, from twin rotors δtr = transition coordinator command e = setpoint smoothness η = setpoint sampling frequency, Hz θc = pitch setpoint command, deg φ, θ, ψ = Euler angles, deg φv, θv, ψv = vertical Euler angles, deg

Online PID Self-Tuning using an Evolutionary Swarm Algorithm with Experimental Quadrotor Flight Results

PID is one of the most common control laws in existence, used extensively across almost every engineering discipline. One potential application of PID is in unmanned air vehicle (UAV) flight. A common problem faced by control law designers is the tuning of such controllers which can be a very time consuming and sometimes difficult process especially on hardware. Various past techniques have considered how PID can be self-tuned. For example, genetic algorithms have been shown to perform offline optimization, which is normally a very long and unviable procedure. In this research we consider the use of online optimization to allow self-tuning of PID. This is done using an ABC colony technique which, compared to genetic algorithms, offers greater simplicity and excellent optimization performance. Applying the online technique has shown to be a difficult task due to the lack of reliability of the online fitness function; the complexity is even more exacerbated in a real-life UAV with a complex PID control architecture. Some important enhancements to the online ABC tuning (in particular the fitness function) have been developed in order to obtain better PID constants. Results on a motion-capture testbed demonstrate that the algorithm combined with the improved fitness function has the ability to select optimal P, I and D values on a quadrotor in real-time; this has never been achieved before experimentally.

Bounded gain-scheduled LQR satellite control using a tilted wheel

Generation of control torque for highly agile satellite missions is generally achieved with momentum exchange devices, such as reaction wheels and control moment gyros (CMGs) with high slew maneuverability. However, the generation of a high control torque from the respective actuators requires high power and thus a large mass. The work presented here proposes a novel type of control actuator that generates torques in all three principal axes of a rigid satellite using only a spinning wheel and a simple tilt mechanism. This newly proposed actuator has several distinct advantages including less mass and more simplicity than a conventional CMG and no singularities being experienced during nominal wheel operation. A new high performance bounded (HPB) linear quadratic regulator (LQR) control law has been presented that extends classical LQR by providing faster settling times, gain-scheduling the control input weightings to optimize its performance, and has much quicker computation times than classical LQR. This work derives a fundamental mathematical model of the actuator and demonstrates feasibility by providing three degree of freedom high fidelity simulations for the actuator using both classical LQR and HPB LQR.

High-Fidelity Modeling and Control of a Twin Helicopter Rotor Tailsitter

A rapidly developing research field is tailsitters, aircraft capable of transitioning between horizontal and vertical flight, a premise that supports a diverse range of applications. This paper considers the high-fidelity modeling of a prototype of a small tailsitter that offers propulsive uniqueness from other known tailsitters in its use of twin helicopter rotors. The tailsitter prototype design is initially considered followed by the development of six degree of freedom nonlinear aerodynamic and propulsive models. Linearized forms of the models are also derived. Two different control laws (PID, LQR) are proposed, and simulation results are presented demonstrating feasibility.

Experimental quadrotor flight performance using computationally efficient and recursively feasible linear model predictive control

A new linear model predictive control (MPC) algorithm in a state-space framework is presented based on the fusion of two past MPC control laws: steady-state optimal MPC (SSOMPC) and Laguerre optimal MPC (LOMPC). The new controller, SSLOMPC, is demonstrated to have improved feasibility, tracking performance and computation time than its predecessors. This is verified in both simulation and practical experimentation on a quadrotor unmanned air vehicle in an indoor motion-capture testbed. The performance of the control law is experimentally compared with proportional-integral-derivative (PID) and linear quadratic regulator (LQR) controllers in an unconstrained square manoeuvre. The use of soft control output and hard control input constraints is also examined in single and dual constrained manoeuvres.

Three-Axis Attitude Control of a Satellite in Zero Momentum Mode Using a Tilted Wheel Methodology

Generation of control torque for highly agile satellite missions is generally achieved with momentum exchange devices, such as reaction wheels and control moment gyros (CMGs) with high slew maneuverability. However, the generation of a high control torque from the respective actuators requires high power and thus a large mass. This paper proposes a novel type of control actuator that will generate torques in all three principal axes of a rigid satellite using only a spinning wheel and a simple tilt mechanism. The tilt mechanism will rotate the spin axis of the wheel (tilt the generated angular momentum vector) about two additional axes thereby generating high control torque about the axes orthogonal to the wheel spin axis. Torque will also be generated about the wheel spin axis through the increase or decrease of the wheel angular speed. This newly proposed actuator generates control torque through controlled precession of the spinning wheel while the tilt angle and the tilt rates are computed without the use of the popular pseudo-inverse calculation obtained with CMGs leading to no singularities being experienced during nominal wheel operation. This paper describes the fundamental mathematical dynamic model of the system and numerical simulations are used to demonstrate the agile three-axis attitude control capability that guarantees a highly efficient trade-off between torque capability, mass, and power consumption. The system actuator sizing is based on slew rates of up to two degrees per second.