Design,development,and flight testing of a 70-gram micro quad-cyclocopter

Abstract

This paper discusses the development of a micro air vehicle scale quad-cyclocopter for the purpose of investigating cyclorotor application to low Reynolds number ( $R e$ ) flight. The 70-gram vehicle is the lightest quad-cyclocopter developed to date by an order of magnitude and only the second to achieve forward flight. It utilized two counter-rotating pairs of cyclorotors operating at $\sim 18, 600$ $R e$ to generate thrust and balance the reaction torque. Each cyclorotor had two control parameters, thrust direction and magnitude, giving the quad-cyclocopter eight independent control parameters. Flight tests were conducted to demonstrate several unique maneuvers made possible by the over-actuated system: changing pitch attitude in a point hover and forward translation via thrust vectoring. Data was collected for longitudinal maneuvers to compare forward flight performance when using strictly thrust vectoring for propulsion versus pitching without thrust vectoring. Similar performance was observed between the two modes for the achieved speeds.

Keywords

Cyclocopter cyclorotor cycloidal rotor micro air vehicles advanced vertical flight

Introduction

Modern Uninhabited Aerial Vehicles (UAVs) have been utilized for a myriad of applications in a broad scope of use-cases from recreational to military. Whether flying around a bridge or industrial structure to inspect for wear, hovering over an agricultural field to measure moisture and yield, or swiftly delivering packages to your front door the utility of UAVs is appreciated in a multitude of sectors. These aircraft have been designed at all sizes from the behemoth 16-ton Global Hawk to the diminutive 30 mg Robobee.^1,2 The term Micro Air Vehicle (MAV) was coined in 1997 when the Defence Advanced Research Projects Agency (DARPA) defined such a vehicle as a UAV weighing 100 grams or less (including a 20-gram payload), having endurance of 1 hour, with a range of 10 km, no dimension over 6 in, and a maximum speed between 10 and 20 m/s.³ MAVs in a military context were envisioned to be cheap, compact, man-portable devices that could be rapidly deployed in a variety of situations requiring remote observation or sensing. Because of their size, these vehicles can operate in constrained spaces like buildings, caves, or dense flora that normally prohibit UAV flight. Soldiers on the ground could leverage all the advantages of remote sensing without hauling pounds of extra gear. Small aircraft also have inherently reduced signatures and therefore, are less likely to be detected, which makes them suitable for clandestine operations. One example of a military MAV is the Black Hornet helicopter from FLIR, shown in Figure 1.

Figure 1.

A soldier deploying a FLIR Black Hornet MAV.

The advantages of an aerial platform at this scale are attractive to not only the military, but also to the broader commercial and personal user. Small platforms can maneuver in confined spaces like industrial complexes or other infrastructure to perform inspections or take photographs while reducing the risk of damage from collisions. Moreover, the reduced signature is beneficial to many users who might not wish to disturb their subjects with the normally irritating noise of larger UAVs (e.g., wildlife or wedding photographers).

However, even though many MAVs have been designed, their adoption and use in military and commercial spaces remains limited despite great improvements in electronics, materials, and battery technology. This fact is related predominantly to two fundamental challenges that arise when designing MAVs, which are (1) the poor performance of airfoils at low Reynolds numbers ( $R e$ ), and (2) the susceptibility to disturbances due to the disadvantageous scaling of physical properties.^4–8 Combined, these physics-related issues have hindered general MAV performance to the degree that MAVs are largely relegated to either toys or research projects.

Challenge - low relative lift at low Reynolds numbers

As $R e$ decreases, the proportion of the shear layer on an airfoil that is laminar will increase. Consequently, the boundary layer will be more likely to separate, thereby reducing the maximum lift and increasing the profile drag because of the change in airflow over the surface, reduced circulation, and increased adverse pressure gradient.⁹ Conventional airfoils are designed to operate at or above a particular $R e$ in order to maximize lift-to-drag ratio, or $C_{l} / C_{d}$ . Below a critical Reynolds number, $C_{l} / C_{d}$ will rapidly plummet.¹⁰ When generalized, this trend can be plotted for various types of airfoils, shown in Figure 2 by the hashed regions.^11–14 In the regime where MAVs operate, even the best performing airfoils fall short of those at higher $R e$ limiting the figure of merit of conventional rotor designs to around 0.65.^6,7

Figure 2.

Effect of Reynolds number on airfoil maximum sectional lift-to-drag ratio.^11,13,15

Challenge - high susceptibility to disturbance at small scales

Disturbance tolerance is detrimentally impacted by the way in which physical properties change with size. An intuitive understanding of this can be obtained from simplified length-based scaling relations. Table 1 shows several physical properties and how each changes with a characteristic length, $L$ . With a halving of scale, the surface area is reduced by $1 / 4$ , mass is reduced by $1 / 8$ , and rotational inertia is reduced by $1 / 32$ . Because mass and rotational inertia decrease much faster than the surface area, the aerodynamic forces at smaller scales have a much greater impact on vehicle motion and therefore, MAVs are more susceptible to wind gusts and external perturbations rendering them incapable of flying except in calm wind conditions.¹⁶ However, for the same reason, the actuators and control surfaces are more effective at smaller scales with higher thrust-to-weight ratios, which makes these small aircraft more agile in quiescent air.^17–19

Table 1.

Relationships governing scaling based on characteristic length $L$ .

Property	Relation
Length	$L$
Surface Area	$L^{2}$
Mass	$L^{3}$
Inertia	$L^{5}$

The novel cyclocopter

In the endeavor to address both of these shortcomings, researchers and engineers have designed all types of MAVs; however, they can generally be divided into three archetypes: fixed-wing, rotary-wing, and flapping-wing. Each method of lift production lends itself to different mission profiles because of its innate strengths and weaknesses. Fixed-wing UAVs tend to have higher endurance while rotary-wing vehicles are more maneuverable. Flapping-wing UAVs are still predominantly experimental due to the mechanical complexity associated with reciprocating wings and the incomplete understanding of flapping flight.²⁰

But the currently limited application of MAV platforms alludes to the need for a non-traditional solution. One possibility is the cycloidal rotor (or cyclorotor), a novel rotary-wing lift-generation device with several unique properties that could fundamentally address the challenges of flight at MAV scales. Unlike a traditional rotor, the cyclorotor utilizes a horizontal axis of rotation with the blade span parallel to the axis. Lift is generated by pitching the blades as they rotate about the azimuth with a cyclic frequency of once per revolution. Figure 3 illustrates the lift and drag on the blades at various points in the rotation along with the total thrust force. Net thrust is upwards because there is a positive geometric angle of attack in the top and bottom halves of the trajectory. Magnitude and direction of the resultant thrust can be changed by altering the amplitude and phase of the cyclic blade pitch kinematics, respectively. Pitch phase is the angle between the azimuth ( $Ψ$ ) of maximum geometric blade pitch and the vertical axis. It is closely linked to the direction of thrust ( $Φ$ ), but with an offset because of the flow curvature effects. Altering the blade pitch phase alters the direction of thrust in the plane of rotation by a similar amount. Being able to quickly adjust the lift production in this manner gives cyclorotors thrust vectoring capabilities that could potentially increase the maneuverability and gust tolerance of an aircraft, thus addressing the challenge of scale. Furthermore, every location along the blade span experiences the same aerodynamic conditions and therefore can be more easily tailored to a specific operating condition to increase efficiency,²¹ potentially overcoming the challenge of low $R e$ flight. The purpose of this research is not necessarily to create a vehicle adhering to the lofty goals in the initial DARPA MAV definition, but to investigate a non-traditional means of overcoming the physical obstacles that have prevented anyone from meeting those goals by building the first ever micro quad-cyclocopter. Development of such a vehicle demonstrates the feasibility of the quad-cyclocopter configuration at these small scales and provides insight into the efforts required to scale a quad-cyclcopter down to the MAV size.

Figure 3.

Blade kinematics and force on a cyclorotor.

Past cyclorotor studies

The cyclorotor concept was envisioned over a century ago, but has only seen limited application since, the most prominent example being the Voith-Schneider Propeller used on aquatic vessels, shown in Figure 4. With the advent of modern materials and electronics, flight-capable cyclorotor-based vehicles, or cyclocopters, have been built at a range of sizes from 30 grams up to over 100 kilograms.^22–24 These were made possible by rigorous scientific study into the fundamental operation and aerodynamics of cyclorotors. Prior research includes extensive analysis conducted by the authors^25–31,22 including detailed performance measurements and flow-field analysis via Particle Image Velocimetry (PIV) for the purpose of optimizing the force production and aerodynamic efficiency of an MAV-scale cyclorotor.

Figure 4.

Two Voith-Schneider propellers installed on a ship.

In order to predict the hover performance of a cyclorotor, an aeroelastic model was also developed and validated for operating Reynolds numbers between 80,000 and 200,000.^32,33 Further, this model was incorporated into a coupled trim model to predict and analyze achievable trim states for twin-cyclocopter designs.³⁴ Note that all the previous cyclorotor performance models assumed 2D aerodynamics because of the high aspect ratio of the blades, were developed for moderate to high Reynolds numbers, and the aeroelastic model approximated the blade structure as a slender beam. However, the current quad-cyclocopter design utilizes low aspect ratio insect-like wings where the aerodynamics are highly three dimensional,²² operates at a Re = 11,000, and uses a plate-like blade structure. Therefore, the previous analytical models could not be directly used for developing the current aircraft; however, an in-house aerodynamic model was used to guide the preliminary design.

The only other documented quad-cyclocopter to achieve forward flight weighed 1 kilogram and had a significantly different cyclorotor design.³⁵ That larger vehicle, developed at the University of Maryland, was capable of flying, rolling on the ground like a car by using its cyclorotors as wheels, and traversing water on floats. It incorporated cyclorotors with blades supported on both ends that had a rectangular planform and thick, symmetrical airfoils. More details about the differences between that cyclorotor design and the one presented in this paper can be found in Refs. 22 and 36.

The subject of this paper is part of a larger body of research by the authors into MAV-scale cyclocopters and is one of several flight-capable cyclocopters developed by the authors, pictured in Figure 5. The 30-gram vehicle is the smallest cyclocopter ever developed with two co-rotating cyclorotors measuring 1 inch in radius.³⁷ Flowfield measurements were taken using PIV on the micro-cyclorotor, which showed that the flow was highly 3-dimensional and vortical,²² unlike many of the previous designs that had largely 2-dimensional flow because of the high aspect ratio rectangular planform blades.²⁸ More information on the aerodynamic conditions of the micro-cyclorotor can be found in Ref. 22. The 33-gram coaxial-nose twin-cyclocopter is the only documented twin-cyclocopter to use a coaxial nose rotor and was used to experimentally extract a linear model of its bare airframe dynamics through system identification techniques. Swapping the single nose rotor for a coaxial rotor system greatly increased handling. Additionally, the linear model and flight-testing showed that the vehicle was passively stable in roll, making it the only cyclocopter to exhibit this property.³⁸ The primary benefit of the coaxial nose rotor was balancing the reaction torque from the nose rotor about the vertical axis such that the vehicle only experienced a net angular momentum about the lateral axis from the co-rotating cyclorotors. Gyroscopic coupling between vehicle states was still present, but to a lesser degree. Motion in one state would not affect the other states as much. The marked improvement in flight quality from balancing one of the two directions of angular momenta present on the body was one impetus behind the development of the quad-cyclocopter, where there is zero net angular momentum on the vehicle and hence no gyroscopic couplings. Additional motivation was the improved actuation potential of a quad-cyclocopter, as will be explained later in the paper. The remainder of this paper discusses the design, development, and flight-testing of a micro quad-cyclocopter to prove viability at MAV-scales and to experimentally investigate its flight dynamics and controls capabilities.

Figure 5.

Hover-capable cyclocopters developed by the authors.^37,38

Quad-cyclocopter configuration

There are two key characteristics that distinguish the quad-cyclocopter from the previously designed twin-cyclocopters. One of which is the presence of additional control parameters giving the quad-cyclocopter greater actuation potential. Each cyclorotor has two actuators: a motor to command RPM and a servo to command the phase of blade pitching (i.e., thrust direction). Commanding pitch phase is similar to commanding thrust direction, but with an RPM-dependent offset caused by rotational effects. The coaxial-nose twin-cyclocopter had two cyclorotors and two nose rotors for a total of 6 control parameters (4 motors and 2 servos). On the quad, four cyclorotors equate to 8 actuators (4 motors and 4 servos). Over-actuation enables the twin and quad to perform longitudinal translation at a constant pitch attitude, not possible for traditional rotorcraft. The additional controls of the quad permit it several more unique maneuvers for heightened mobility, namely point hover within a range of pitch attitudes and lateral translation at a constant body attitude. The specifics of these maneuvers will be discussed in a later section.

The second characteristic is the balancing of all angular momenta on the body. Nose rotors on the twin-cyclocopter serve the dual purpose of countering the cyclorotor reaction moment and providing extra lift; however, counter-rotating cyclorotors could perform the same functions instead. There are many cyclocopter configurations that can lead to zero total angular momentum — using multiple cyclorotors of the same size rotating at the same speed or several sizes of cyclorotors at different speeds. From the possible options, only those that used one size of cyclorotor were considered for this study to avoid the need of developing two different sizes of cyclorotors for a single vehicle.

Available quad-cyclorotor configurations were then limited to two options, either a “+” or an “H”. In the “+” configuration (Figure 6) the opposing cyclorotors are contra-rotating. While in hover or low-speed flight this creates a highly mobile vehicle capable of longitudinal and lateral translation with no change in body attitude. However, with respect to oncoming air flow in forward flight, one cyclorotor is forward-spinning while the other is back-spinning, which causes an increasing thrust differential as advance ratio increases due to the virtual camber effect, which has a similar impact as that of advancing and retreating sides of a conventional helicopter rotor. For more information on the virtual camber effect see Ref. 39. The “H” configuration (Figure 7) sacrifices lateral translation ability for improved forward flight speed and performance because there is no lift differential in forward flight. Opposite cyclorotors now spin in the same direction while the front and back pairs counter-rotate. Rotation is such that the outermost blades are descending and the innermost blades are ascending, which is necessary to enhance the pitch control moment.

Figure 6.

Quad-cyclocopter utilizing a “+” cyclorotor arrangement.⁴⁰

Figure 7.

Quad-cyclocopter utilizing an “H” cyclorotor arrangement.⁴¹

To explain why, consider a nose-up pitch maneuver. A pitch-up moment is generated by increasing the RPM of the front cyclorotors, while that of the rear cyclorotors is decreased. Not only does this change the cyclorotor thrust, but it also imparts a torque on the body due to the change in power required to spin the motor. When the cyclorotors are spun towards the center (Figure 8), the torque change ( $Δ$ Q) resulting from the increase to front RPM ( $Ω$ ) and decrease to rear RPM results in a nose-down pitching moment ( $Δ$ Q $_{n e t}$ ), opposite to the intended motion. If the torque magnitude is high enough, the vehicle may even briefly pitch down before pitching up. By spinning the cyclorotors away from the center (Figure 9), reaction torque is now in the desired pitching direction and works in concert with the thrust differential ( $Δ$ T $_{n e t}$ ) to pitch the vehicle.⁴² Alternatively, if carefully designed, all four cyclorotors can spin in the same direction and the CG can be placed appropriately to counteract all pitching moments as demonstrated by CycloTech GmbH.⁴³ Forward flight would then increase lift equally on all four cyclorotors; however, pitch attitude then becomes integral to trim.

Figure 8.

Inward rotation of cyclorotors with opposing reaction torque and lift differential pitching moment.

Figure 9.

Outward rotation of cyclorotors with cooperating reaction torque and lift differential pitching moment.

Ultimately, the counter-rotating “H” configuration was chosen for the present design because of the increased forward flight performance over the “+” arrangement. This configuration also enabled the ability to freely change pitch attitudes, which is not allowed by the co-rotating twin-cyclorotor design. Development began with a survey of available commercial off-the-shelf (COTS) parts followed by an estimation of the total vehicle weight from the selected hardware. By mirroring the previously designed twin-cyclocopter about its lateral mid-line a quad-cyclocopter could be created. Estimated weight of a quad-cyclocopter produced in this manner was calculated to be 55 grams; however, each cyclorotor was only designed to produce 10 grams. A detailed discussion of the optimized cyclorotor utilized on the twin-cyclocopter can be found in Refs. 22,37. For the quad-cyclocopter, new cyclorotors capable of more thrust had to be developed.

Cyclorotor development

Rather than completely redo the design process carried out to optimize the cyclorotor for the twin-cyclocopter,²² it was assumed that the increase in Reynolds number would be small enough to not alter the design parameters that had been experimentally optimized. As a result, a scaling approach was taken to geometrically scale the cyclorotor from the previously optimized twin-cyclocopter by assuming a $L^{2}$ relationship between size and thrust. To lift the estimated 55-gram weight each cyclorotor had to generate at least 13 grams of thrust, hence a requirement of 16 grams was chosen to allow for some margin. Going by the length-based scaling laws mentioned previously, increasing cyclorotor size 25% would meet the 16-gram requirement, up from the previous design that produced just over 10 grams of thrust. For some additional margin of safety, an increase of 30% was decided upon from which a 69% gain in thrust would be expected for a maximum total thrust of 68 grams ( $4$ cyclorotors $\times 10$ grams $\times {1.3}^{2}$ ). Scaling the 1-inch cyclorotor used on the twin-cyclocopter by 30% established a radius of 1.3 inches for the cyclorotors to be used on the quad. An operating speed of 4000 RPM was maintained for a commensurate increase in Reynolds number from $\sim 11, 000$ to $\sim 18, 600$ (calculated at 75% span). The following parameters were maintained from the previous 1-inch design: 1.62 aspect ratio blades with an elliptical planform, flat plate airfoils, 0.8 $c h o r d / r a d i u s$ ratio, and $\pm 45^{\circ}$ symmetric pitching. A 0.8 $c h o r d / r a d i u s$ ratio at the blade root gives a root chord length of 1.04 inches for a span of 1.68 inches. A 30% larger cyclorotor with similar geometric parameters operating at the same RPM should produce approximately 69% more thrust. For more details on the previous cyclorotor optimization study see Ref. 22.

Cyclorotor and pitching mechanism design

The cyclorotors used a cantilevered blade design that leveraged the inherent increase in structural stiffness-to-mass ratio associated with reducing scale to produce a lightweight, yet stiff cyclorotor. A single end-plate made of two $1 / 32 "$ carbon fiber laminate frames separated by Delrin® spacers supported the centrifugal and aerodynamic loads from the blades. Ensuring stiffness was important because the blade deflections could reduce thrust by up to 40 percent.⁴⁴ Figure 10 shows the cyclorotor design with the frames and spacers labeled, as well as the non-rotating pitch offset in the middle that was necessary to achieve cyclic pitching as part of the 4-bar mechanism, explained below. Two ball bearings inside the spacer created a cantilevered root condition for the blade while allowing it to rotate freely. A PEEK (polyetheretherketone) bushing between the ball bearings and the 0.7 mm carbon fiber rod blade root properly centered the blades and provided a consistent interface between the blades and the ball bearings. The protrusions on the front frame protected blade leading edges in the event that the vehicle contacted an obstacle.

Figure 10.

$1.3 "$ radius cyclorotor with key components labeled.

A passive 4-bar linkage mechanism was used to cyclically pitch the blades. Figure 11 shows a schematic of the mechanism and the four lengths that comprise it. On the vehicle, this system was sandwiched between the two frames for protection and reduction of parasitic drag. $L_{1}$ is the cyclorotor radius, $L_{2}$ is the length of the pitch offset, $L_{3}$ is the pitch link length, and $L_{4}$ is the distance between the blade rotation axis and the pitch link connection point. $L_{1}$ and $L_{4}$ are fixed as constraints, which made $L_{2}$ and $L_{3}$ the design parameters determined by maximum geometric pitch attained at the top and bottom of the circular trajectory — in this case $+ 45^{\circ}$ at the top and $- 45^{\circ}$ at the bottom, or $\pm 45^{\circ}$ symmetric pitching. If the angle at the top was different versus the angle at the bottom it would be referred to as asymmetric pitching. Table 2 contains the lengths used in the 4-bar mechanism.

Figure 11.

4-bar linkage system implemented on the micro-cyclorotor.⁴⁵

Table 2.

Lengths used in the 4-bar pitching mechanism.

Linkage	Length (in)
$L_{1}$ , cyclorotor radius	1.3
$L_{2}$ , pitch offset	0.368
$L_{3}$ , pitch link	1.351
$L_{4}$ , blade root	0.52

Figure 12 shows an exploded view of the cyclorotor and pitching mechanism. Pitch links (B) were manufactured from a single strand of unidirectional carbon fiber prepreg laid into a Teflon^TM mold with a silicone mat to compress the fibers into the mold. Using a silicone mat instead of a male half of the mold evenly distributed the clamping force and relieved pressure points that could damage the carbon fibers. This mold-layup process is similar to the method used to manufacture the blades described in the next section. The center-to-center distance in the mold between the loops at the ends of the pitch links was made 0.018 inches shorter than the designed length to account for some post-removal stretch of the linkage. Pitch links made in this fashion weighed 0.01 grams and were strong in the direction of loading because of the alignment of the fibers with the load. Loops at either end of the pitch links fit around bushings on the other components for ease of movement. At the center of the cyclorotor a Delrin® peg glued into the pitch offset (C) served as the bushing for all four pitch links. On the blades, a PEEK bushing (G) was fixed onto a 0.7 mm carbon fiber rod at the trailing edge.

Figure 12.

Exploded view of the cyclorotor, pitching mechanism, and blades with parts labeled.²²

Blade design and manufacturing

The hallmark of the micro-cyclorotor design for the twin-cyclcopter was its lightweight, elliptical-planform, flat-plate-airfoil blades and a more detailed description of their development can be found in Ref. 22. A brief overview is provided here. The bio-inspired elliptical planform was chosen to improve aerodynamic efficiency while mimicking flying creatures that operate at similar Reynolds numbers. A flat-plate airfoil was chosen because it performs better than conventional airfoils at these low Reynolds numbers.¹⁵ Keeping blade weight low was imperative to creating a lightweight cyclorotor because every unit of blade weight generated 590 times that amount of centrifugal force (acceleration of 590 g), causing greater deflection, in turn driving up cyclorotor structural weight required to resist bending. Each blade weighed approximately 0.12 grams which corresponded to 71 grams of centrifugal force, well above the 16 grams of aerodynamic loading. In order to manufacture lightweight blades with consistent properties and quality, a Teflon^TM mold layup similar to that used for the pitch links was employed (shown in Figure 13). First, an extra-long piece of unidirectional carbon fiber prepreg was placed into a channel in the Teflon^TM mold that formed the mid chord such that some length protruded beyond the top of the groove for the planform shape. Then the planform shape of the blades was formed by laying a continuous length of unidirectional prepreg in the this groove to make two passes around the edge. Between the first and second passes, a length of 0.7 mm unidirectional carbon fiber rod was placed at the mid chord and trailing edge to act as the pitching axis and pitch link attachment point, respectively, shown in Figure 13. Placing the blade pitching axis at the mid chord (i.e., coincident with the blade center of gravity) through structural design reduced blade elastic twist from centrifugal forces. Next, the extra length of prepreg protruding beyond the top of the planform groove at the mid chord was folded back over onto the carbon fiber rod sandwiching the two outer passes and rod. Doing so increased contact surface area between the layers and improved bond strength. Finally, the Teflon mold and silicone mat were clamped between sheets of aluminum and placed in an oven to cure for 135 minutes at $350^{\circ}$ F. Once removed from the mold and cleaned up, a 3 micron thick Mylar sheet was affixed to both sides with spray-on contact cement. Item D in Figure 12 shows the blade structure after it has been removed from the mold, but before Mylar has been added.

Figure 13.

Mold layup assembled in C-clamp prior to curing (offset to show layers).

Vehicle design

With the cyclorotors designed, they could then be incorporated into an “H” configuration quad-cyclocopter using a combination of custom and commercial off-the-shelf (COTS) parts. The 58-gram quad-cyclocopter shown in Figure 5 was the first iteration of the vehicle with power supplied at 3.85 V by two 260 mAH LiHV (Lithium polymer High-Voltage) batteries wired in parallel (1 cell in series, or 1S) on the underside of the airframe. Although this vehicle could lift-off and hover, flight time was severely restricted because the voltage drop under load reduced the motor RPM. To compensate for this quick voltage drop, the cyclorotor power system was converted to 7.7 V 2S (2 cells in series). The under-slung batteries were rewired in series and the electronic speed controllers (ESCs) were upgraded to handle the additional voltage. A second 260 mAH LiHV battery was added on top of the airframe to supply 1s power to the flight controller and servos. These changes added weight, bringing the total to 70 grams, but allowed the motors to spin the cyclorotors at a higher RPM in order to produce more than the initial design of 68 grams total thrust. Maximum thrust with the new electronics was not measured because the vibrations on the constrained test stand at higher RPMs threatened the structural integrity of the vehicle. Regardless, the additional voltage and thrust allowed the quad-cyclocopter to fly for approximately 3 minutes while performing a mix of hover and translational flight.

Selective laser sintering (SLS) was used to manufacture the airframe from glass-filled nylon with carbon fiber reinforcement. It was designed to hold all of the components rigidly in a compact arrangement, shown in Figure 14 with the top structure removed for clarity. The stationary main shaft of the cyclorotors is directly aligned with the rotational output of the servos and the cyclorotor is driven around it by a Hobbyking AP-03 7000 kV motor via a 5.45:1 single-stage gear reduction. The DYS XSD7A electronic speed controllers (ESCs) are glued directly to the airframe and the flight controller is soft-mounted at the CG using foam tape. The servos used were Hobbyking HK-5320 digital servos that were lightened to 1.25 grams each from their factory weight of 1.7 g by removing excess casing, filing down the remaining casing, and cutting off excess wiring. Figure 15 shows the final 70-gram configuration of the vehicle which measures $6.6 " L \times 7.7 " W \times 4.4 " H$ not including the landing gear. See Table 3 for a weight breakdown by sub-system.

Figure 14.

Top view of the quad-cyclocopter with components labeled showing arrangement on the airframe.

Figure 15.

70-gram quad-cyclocopter.

Table 3.

Sub-system weight breakdown of 70-gram quad-cyclocopter.

Component	Weight (g)	Total ( $%$ )
Motors + Transmission	13.4	19
Digital Servos	5	7
Cyclorotors	14.4	21
Structure + Wires	13.1	19
LiHV Batteries	18.4	26
Electronics	5.7	8
Total	70	100

Vehicle telemetry

The flight controller was a custom-built, embedded processor-sensor board called ELKA-R that was developed at the University of Maryland to facilitate the research of micro air vehicles⁴⁶. It can be seen next to a US quarter in Figure 16 and also in Figure 5 mounted on the cyclocopter at its center of gravity. Total weight of the board was 1.7 grams and it was powered by a single 1-cell 3.85-volt 260 mAh LiHV battery. The flight controller housed a STM32 microprocessor with a 32-bit ARM Cortex F4 core for onboard computational tasks. The MPU-9150 inertial measurement unit (IMU) integrated on the board includes tri-axial gyroscopes, tri-axial accelerometers, and magnetometers. Wireless communications were serviced by an on-board nRF24L01 chip, a low-power 2.4 GHz RF transceiver. The flight controller had a sensor update rate of 500 Hz. Communication with the base station occurred at 50 Hz, which was utilized for streaming vehicle attitude, actuator controls data, and receiving pilot commands⁴⁷. The flight controller was capable of sensing vehicle attitude angles and angular rates and sending corrective signals to the servos for stabilization by varying the pulse-width input to the motors and servos.

Figure 16.

Custom-built 1.7-gram kinematic autopilot with U.S. quarter for size comparison.

To communicate wirelessly with the onboard controller, the operator used a LabVIEW interface, which included a wireless 2.4 GHz data link with nRF24L01 transceiver. The base station LabVIEW program allowed the operator to control the vehicle, modify feedback gains, change the sensitivity of pilot inputs, and record attitude data transmitted by the onboard processor. The LabVIEW program received pilot inputs through the use of a DX6i Spectrum transmitter which was hardwired to the base station. The program then communicated with the microcontroller through a wireless radio link and used this connection to send the control inputs and receive the vehicle attitude angle and angular-rate data. The data processing and inner-loop feedback control calculations were performed onboard by the microprocessor.

The on-board gyros measured the pitch ( $q$ ), roll ( $p$ ), and yaw ( $r$ ) angular rates while the accelerometers recorded the tilt of the gravity vector in the body frame. The vehicle attitude could then be extracted by integrating the gyro measurements with time. However, it was known that this would lead to drift in attitude measurements.⁴⁸ Accelerometers on the other hand offered stable bias, but were sensitive to vibrations and in general offered poor high frequency information.⁴⁹ Therefore, a complementary filter was incorporated to extract the pitch and roll Euler angles using a high-pass filter for the gyros and a low-pass filter for accelerometers, both with a 10 Hz cut-off, which was experimentally derived from flight-testing the quad-cyclocopter. This value was found to reduce noise while permitting sufficient responsiveness in the controller to perturbations and commands. The cyclorotor vibrations were filtered out since they are sufficiently higher than the body dynamics. On-board inner-loop feedback was implemented using a cascaded proportional-integral-derivative (PID) controller, shown in Figure 17, that had an inner PID loop on the attitude rates ( $p$ , $q$ , and $r$ ) and and outer loop with only proportional gains on pitch and roll Euler angles ( $θ$ and $ϕ$ ) — a PPID loop. Heading measurements had not been implemented on the ELKA-R at this point and therefore, yaw Euler angle was not measured or used for feedback.

Figure 17.

PPID feedback loop architecture for attitude stabilization and control.

Attitude control

Thrust magnitude was modulated by changing motor RPMs while thrust direction could be varied by rotating the pitch offset links using the servos. Both of which could be commanded independently for each of the four cyclorotors, for a total of eight control parameters. Through combinations of magnitude and phase control, the aircraft could be commanded to roll, pitch, and yaw along with several other unique maneuvers afforded to it by the over-actuated nature. A roll moment was generated by increasing motor RPMs on one side and reducing them on the other to create a thrust differential (Figure 18). Similarly, a differential between fore and aft motor RPMs generated a pitching moment because of the change in thrust and torque (Figure 19). Yaw was controlled by tilting the thrust vectoring servos forward on one side and backward on the other (Figure 20).

Figure 18.

Differential RPM generating a positive rolling moment.

Figure 19.

Differential RPM and torque being used to create a positive pitching moment.

Figure 20.

Differential thrust vectoring being used to generate a positive yawing moment.

Angling all thrust vectors in the same direction produced a pure longitudinal translation with no change in pitch attitude (Figure 21), a capability shared between the twin- and quad-cyclocopters. Specific to the quad-cyclocopter was the ability to perform a point hover within a range of different pitch attitudes. Figure 22 shows how the thrust vectors were skewed simultaneously with a change in pitch to achieve a hovering trim at a non-zero body attitude only limited by the servos’ range of motion. A trim setting was included in the controller that allowed the pilot to set a desired pitch attitude in hover.

Figure 21.

Simultaneous thrust vectoring being used to generate a longitudinal force.

Figure 22.

Flight test demonstrating hover in a non-zero body attitude achieved via thrust vectoring.

Another possible maneuver not explored in this work is a pure lateral translation, which would have required a modification of the quad-cyclocopter design. If the front co-rotating cyclorotors are swept in one direction and the rear cyclorotors are swept in the opposite direction the servos could be used to produce a pure lateral force on the body by pointing the thrust vectors towards each other on one side of the body and away on the other side. Figure 23 shows an illustration of such an arrangement (n.b., the motors and airframe were not modified to accommodate the swept cyclorotors). Longitudinal forces would be canceled out and only a side force in one direction would be imparted. Maximum side force amplitude would be determined by sweep angle of the cyclorotors at the cost of forward flight performance, whereas a $45^{\circ}$ sweep angle would produce a modified “+” configuration, or an “x” arrangement quad-cyclocopter. The difference being which cyclorotors are used to command pitch and roll. For a “+” configuration, only a single opposing pair of cyclorotors is used to generate a moment about the longitudinal or lateral body axis. In an “x” configuration, a pitch or roll moment is generated by utilizing both opposing pairs of cyclorotors, employing all four. Cyclorotors in an “x” configuration would partially alleviate the asymmetric thrust problems of the “+” configuration in forward flight because cyclorotors on either side of the vehicle are spinning in the same direction, which would not produce a rolling effect due to a differential advance ratio on one side versus the other, but this potential remains open for future study.

Figure 23.

Top-down view of lateral force production by cyclorotors with $15^{\circ}$ sweep angle.

Flight testing

Making the quad-cyclocopter flight-ready was a process comprised of four steps (1) trim the servo angles or positions (2) trim the motor speeds (3) tune the rate-based inner PID loop, and (4) tune the outer proportional gains on attitude angles. Trimming the servo angles was done by using a single-axis yaw stand to adjust all the thrust vectors until they were perfectly vertical (Figure 24). Each cyclorotor was spun up individually to a set RPM and then servo position was altered until no motion about the yaw axis was observed. Afterwards, hopping flights were performed without feedback to adjust motor RPMs until roll and pitch biases caused by minor differences in the hardware were removed. Initial flights exhibited moderate stability in yaw; therefore, yaw damping had to be improved via increasing the derivative feedback gains. Once vertical ascension was confirmed, controller gains could be added successively to the inner rate-based PID loop. Properly tuned, the vehicle could be flown using only feedback on angular rates, but with large amounts of drift and high pilot workload. The outer loop proportional gains on roll and pitch attitude greatly reduced both. Time required to trim and tune the controller in order to achieve stable hover for the quad-cyclocopter was comparatively lower relative to the twin-cyclocopters because of the reduced complexity; no consideration had to be given to the gyroscopic couplings, controls mixing, or feedforward gains associated with twin-cyclocopter flight caused by the nose rotor.³⁸ Flight tests were performed to analyze the vehicle’s response in roll, pitch, yaw, and heave. For all except pitch, behavior similar to that of a typical quadcopter was observed, which is to be expected considering the similarity in force distribution and control strategies employed. Novel behavior was observed in pitch and unique maneuvers in the longitudinal mode were performed that are only possible because of the cyclorotor’s thrust vectoring capability.

Figure 24.

Micro quad-cyclocopter mounted on a single degree of freedom yaw stand.

Longitudinal mode

The transmitter pitch stick was configured to command longitudinal motion using either pitch attitude or by thrust vectoring using all of the pitch phase control servos with a switch to toggle between the two modes that could be flipped mid-flight. Figure 25 shows the vehicle in forward flight with a commanded thrust vector phase angle ( $Φ$ ) while maintaining a level attitude ( $θ$ ). Longitudinal translation stick inputs were fed directly to the phase angle servos in a feedforward manner with no feedback on position, longitudinal velocity, or longitudinal acceleration. Alternatively, Figure 26 shows an image of the quad-cyclocopter in forward flight attained through a non-zero pitch attitude angle commanded by differential cyclorotor torque while thrust vector phase remained zero. A video demonstration of both flight modes can be found in Ref. 50.

Figure 25.

Thrust vectoring being used to translate with a level attitude (motion is left to right).

Figure 26.

Pitch attitude utilized for forward flight (motion is right to left).

Flight data from a characteristic set of longitudinal maneuvers emphasizes the differences between thrust vectoring and pitching motion. Figure 27 shows body frame longitudinal velocity, $u$ , achieved using both methods during straight and level flight. From the start of the data to about 14.5 seconds, the flight controller maintained a constant pitch attitude while the vehicle was commanded to translate using thrust vectoring, which can be seen in Figure 28. Figure 29 plots how thrust vector phase angle was modulated from high to low (fore and aft) in a stepwise fashion to excite a corresponding velocity. At approximately 14.5 seconds the switch on the Spektrum DX-6i transmitter was flipped to change modes from thrust vectoring to pitching, indicated by the maroon vertical line at the 14.5 second mark. Thrust vector phase then remained constant, as seen from Figure 29 after the vertical line, while the flight controller commanded a non-zero pitch attitude, as shown after the vertical line in Figure 28, to create longitudinal motion.

Figure 27.

Flight data of longitudinal velocity, $u$ , achieved with two different flight modes.

Figure 28.

Flight data of pitch attitude angle, $θ$ , during operation under two different flight modes.

Figure 29.

Flight data showing the commanded thrust vector phase angle during operation under two different flight modes.

During thrust vectoring flight, maximum thrust vector phase angle, $Φ$ , was $- 14^{\circ}$ (pointing backwards) at 11 seconds with a corresponding maximum $- 1.4$ m/s velocity. In pitching mode, maximum pitch attitude, $θ$ , was $- 20^{\circ}$ (nose down) which resulted in a maximum velocity of 1.7 m/s at 16 seconds. In both cases, the vehicle was necessarily being propelled by a longitudinal component of the angled maximum thrust; therefore, a larger angle resulted in a higher maximum velocity. This is also evident in the acceleration of the vehicle, which is higher for the pitching maneuvers that reach a higher speed in a shorter time.

By combining a pitch command with an equivalent converse thrust phase command the quad-cyclocopter can change its pitch attitude while remaining in a point hover. Figure 30 shows the data from such a maneuver collected in flight at two different frequencies of manually-controlled inputs demonstrating near-zero $u$ drift while pitch angle oscillates $\pm 15^{\circ}$ . An empirically derived calibration constant converted the angle command into the corresponding pulse width modulation command for the servos. Small amounts of longitudinal velocity (approximately $\pm 0.20$ m/s) are created during the maneuver primarily due to a small amount of error in the calibration constant and, to some extent, as a result of the different actuation speeds.

Figure 30.

Flight test data showing pitch attitude modulation in stationary hover.

These results reveal a critical aspect of the quad-cyclocopter MAV, that equivalent forward flight performance can be achieved using either thrust vectoring or pitching. The independent nature of these two control methods allows them to be used separately or in conjunction based on the intended purpose. For instance, fixed wings could be added to enhance performance or extend range while pitch attitude is augmented to adjust the wing’s angle of attack separately from thrust direction. Furthermore, a payload attached to the body could be pointed by pitching the vehicle because any attitude about the lateral axis is achievable, only being limited by the actuation range of the servos (e.g., $360^{\circ}$ servos would enable the quad-cyclocopter to do a complete flip). Unlike the traditional quadcopter that cannot vector its thrust, any attitude can be maintained while in forward motion and because the quad-cyclocopter does not need to constantly change body attitude to fly the need for a heavy active gimbal mount is reduced. One of the body axes can now be stabilized independently from body motion, which means a payload gimbal mount requires fewer axes and potentially that digital stabilization alone might be sufficient. Simplifying or eliminating the mounting hardware for payloads could compensate for the additional mechanical complexity of a quad-cyclocopter relative to a conventional quadcopter. Figure 22 shows an image of the vehicle hovering at a $15^{\circ}$ attitude after taking off at a level attitude, which demonstrates this capability.

However, in order to fully harness the capabilities of the quad-cyclocopter there are several considerations that need to taken into account. Firstly, in high-speed forward flight or at high pitch attitude the effects of drag and cyclorotor interaction need to be considered. High pitch angles will create more drag on the body; however, high-speed flight at level attitude might cause adverse interactions between the front and rear cyclorotors. Offsetting these effects to reach maximum speed could involve finding an optimal pitch attitude that is dependent on forward velocity to minimize the total impact of both. Design changes could also mitigate the issue entirely by laterally or vertically staggering the cyclorotors. Secondly, thrust vectoring and pitch attitude control methods have different response times. Changing pitch attitude requires overcoming vehicle inertia, whereas thrust vectoring depends on the actuator speed. The disparity between the two approaches will grow with scale as inertia slows down the vehicle dynamics, reducing the response time. At MAV-scales with relatively small inertia the difference between the two control methods is small; however, on larger vehicles the faster response time of thrust vectoring might impart greater maneuverability to a cyclocopter above that of traditional vertical flight aircraft designs.

Summary and conclusions

Cyclorotors are novel lift generation devices that are still in the nascent stages of being understood and utilized in flying vehicles. The first flight capable cyclocopter was developed relatively recently in an effort to study the potential advantages cyclocopter flight might possess related to maneuverability, efficiency, and gust tolerance.^44,45 The research in this paper examines how cyclorotors applied at MAV-scales might compensate for some of the inherent detriments associated with small aircraft, specifically low Reynolds number flight and poor gust tolerance. Presented herein was the design, development, and flight-testing of a 70-gram quad-cyclocopter for the purpose of exploring its unique capabilities. The vehicle used 4 cyclorotors with cantilevered, elliptical planform, flat-plate airfoil blades in an “H” arrangement and was the smallest quad-cyclocopter ever made. A systematic flight-testing procedure was employed to safely trim the 8 independent control parameters to achieve free flight. Subsequently, data was collected on longitudinal maneuvers to compare different modes of translation control. To date, it is one of only two quad-cyclocopters that have achieved forward flight, the other being more than an order of magnitude heavier and a substantially different design. Several important insights were gained through this work:

Eliminating all angular momenta allowed for the simplification of flight control laws by removing the need to compensate for gyroscopic couplings or asymmetrical controls possessed by twin-cyclocopter configurations.

Improved handling and flight quality was observed with the micro quad-cyclocopter relative to the twin-cyclocopter resulting from over-actuation (eight vs six independent control parameters) and lack of gyroscopic coupling.

The independent control parameters on a quad-cyclocopter can be combined in various ways to perform several unique maneuvers atypical of traditional hover-capable MAVs. Pure longitudinal translation at a level body attitude and a point hover at multiple different pitch attitudes were demonstrated.

All active controls were achieved using either motors or servos without mixing the two into one control. Not having multiple actuator types in a single control methodology meant that no allowance had to be made for differences in latency or control authority. Hover attitude modulation used both motors and servos in concert as a pitch trim state adjustment, not an active control.

While the vehicle developed in this work is not ready for military or commercial use at this point, it was a necessary first step towards proving the viability of the cyclorotor concept and the quad-cyclocopter configuration at such small scales. The unique maneuvers that the quad-cyclocopter can perform could potentially be used to increase gust tolerance relative to traditional quadcopters and the simplicity of controller design and the absence of oscillatory inertial loads from flapping wings makes development more accessible than a flapping-wing MAV. But fully utilizing these advantages requires a deeper understanding of the aerodynamics and flight characteristics of the quad-cyclocopter.

Future work could involve rigorous investigation of the flight dynamics of the quad-cyclocopter in hovering and forward flight using system identification techniques to extract a linear flight dynamics model. Outer-loop, computer-controlled feedback could then be added to command more complex flight paths and maneuvers. This capability could be used to explore the limits of the MAV cyclocopter’s performance and validate any potential benefits over traditional rotorcraft MAVs. A gust generation device could also be used to probe the gust tolerance of the cyclocopter and examine the best way to make use of its over-actuated nature. Aside from that, a more thorough analysis of the engineering trade-offs involved in designing cyclocopters and the various potential configurations is required.

Footnotes

Declaration of conflicting interests

The authors declare that there are no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This research was partially supported by the U.S. Army’s Micro Autonomous Systems and Technology–Collaborative Technology Alliance (MAST-CTA) with Chris Kroninger (Army Research Laboratory–Vertical Technology Directorate) as Technical Monitor.

The work was also partially supported by Army/Navy/NASA’s Vertical Lift Research Center of Excellence (VLRCOE) led by the University of Maryland with Dr. Alex Moodie and Dr. Mahendra Bhagawat as Technical Monitors.

ORCID iD

Carl Runco

References

Chirarattananon

Fuller

et al. Controlled flight of a biologically inspired, insect-scale robot. Science 2013; 340: 603–607.

Jafferis

Helbling

Karpelson

et al. Untethered flight of an insect-sized flapping-wing microscale aerial vehicle. Nature 2019; 570: 491–495.

McMichael

Francis

URCMS

. Micro air vehicles toward a new dimension in flight. Technical Report 13, 2000. Originally published online at http://www.darpa.mil/tto/MAV/mavauvsi.html, Aug. 7, 1997.

Spedding

Lissaman

PBS

. Technical aspects of microscale flight systems. J Avian Biol 1998; 29: 458.

Epstein

. The role of size in the future of aeronautics. Bridge (Kans City) 2004; 34: 17–23.

Pines

Bohorquez

. Challenges facing future micro-air-vehicle development. J Aircr 2006; 43: 290–305.

Chopra

. Hovering micro air vehicles: Challenges and opportunities. In Proceedings of American Helicopter Society Specialists’ Conference. Seoul, Korea: International Forum on Rotorcraft Multidisciplinary Technology.

Galinski

Zbikowski

. Some problems of micro air vehicles development. Bulletin of the Polish Academy of Sciences: Technical Sciences 2007; 55: 91–98.

Anderson

. Introduction to flight. 9 ed. Columbus, OH: McGraw-Hill Education, 2021. ISBN 1260226743.

10.

Selig

Donovan

Fraser

. Airfoils at Low Speeds. 1504 North Horseshoe Circle, Virginia Beach, Virginia 23451: H. A. Stokely, 1989.

11.

McMasters

Henderson

. Low-speed single-element airfoil synthesis. In NASA. Langley Res. Center The Sci. and Technol. of Low Speed and Motorless Flight, Pt. 1. The Science and Technology of Low Speed and Motorless Flight, NASA Langley Research Center.

12.

Lissaman

PBS

. Low-reynolds-number airfoils. Annu Rev Fluid Mech 1983; 15: 223–239.

13.

Mueller

. Aerodynamic measurements at low reynolds numbers for fixed wing micro-air vehicles. Technical report, NOTRE DAME UNIV IN DEPT OF AEROSPACE AND MECHANICAL ENGINEERING, 1999.

14.

Kunz

. Aerodynamics and design for ultra-low Reynolds number flight. PhD Thesis, Stanford University, 2003.

15.

Winslow

Otsuka

Govindarajan

et al. Basic understanding of airfoil characteristics at low reynolds numbers (104–105). J Aircr 2017; 55: 1050–1061.

16.

Zarovy

Costello

Mehta

. Experimental method for studying gust effects on micro rotorcraft. Proc Inst Mech Eng, Part G: J Aerosp Eng 2012; 227: 703–713.

17.

Burk

Wilson

. Radio-controlled model design and testing techniques for stall/spin evaluation of general-aviation aircraft. In SAE National Business Aircraft Meeting.

18.

Mettler

. Identification modeling and characteristics of miniature rotorcraft. New York, NY: Springer, 2003.

19.

Ivler

Rowe

Martin

et al. System identification guidance for multirotor aircraft: Dynamic scaling and test techniques. J Am Helicopter Soc 2021; 66: 1–16.

20.

Galinski

. Influence of mav characteristics on their applications. Aviation 2005; 9: 16–23.

21.

Benedict

Jarugumilli

Chopra

. Experimental optimization of MAV-scale cycloidal rotor performance. J Am Helicopter Soc 2011; 56: 22005–2200511.

22.

Runco

Himmelberg

Benedict

. Experimental studies on a mesoscale cycloidal rotor in hover. J Aircr 2018; 1–10. DOI: 10.2514/1.c034736.

23.

Lee

Min

Lee

et al. Design and experiment of two-rotored uav cyclocopter. In 29th Congress of the International Council of the Aeronautical Sciences. St. Petersburg, Russia. 7–12.

24.

Lee

Min

Lee

et al. Design, analysis, and experimental investigation of a cyclocopter with two rotors. J Aircr 2016; 53: 1527–1537.

25.

Sirohi

Parsons

Chopra

. Hover performance of a cycloidal rotor for a micro air vehicle. J Am Helicopter Soc 2007; 52: 263–279.

26.

Benedict

Ramasamy

Chopra

et al. Experiments on the optimization of the mav-scale cycloidal rotor characteristics towards improving their aerodynamic performance. In Proceedings of the International Specialists’ Meeting on Unmanned Rotorcraft. Scottsdale, Arizona. 20–22.

27.

Benedict

Jarugumilli

Chopra

. Design and development of a hover-capable cyclocopter mav. In Proceedings of the 65th Annual National Forum of the American Helicopter Society. Grapevine, TX. 27–29.

28.

Benedict

Ramasamy

Chopra

et al. Performance of a cycloidal rotor concept for micro air vehicle applications. J Am Helicopter Soc 2010; 55: 22002–2200214.

29.

Benedict

Jarugumilli

Chopra

. Experimental performance optimization of a mav-scale cycloidal rotor. In Proceedings of the AHS Specialists’ Meeting on Aeromechanics. San Francisco, CA, pp. 20–22.

30.

Walther

Coleman

Benedict

. Understanding unsteady aerodynamics of cycloidal rotors in hover at ultra-low reynolds numbers. In 2018 AIAA Aerospace Sciences Meeting. American Institute of Aeronautics and Astronautics. DOI:10.2514/6.2018-0080.

31.

Walther

Saemi

Benedict

et al. Symmetric versus asymmetric pitching of a cycloidal rotor blade at ultra-low reynolds numbers. J Aircr 2019; 56: 1179–1199.

32.

Benedict

Mattaboni

Chopra

et al. Aeroelastic analysis of a micro-air-vehicle-scale cycloidal rotor in hover. AIAA J 2011; 49: 2430–2443.

33.

Halder

Benedict

. Free-wake based nonlinear aeroelastic modeling of cycloidal rotor. In AIAA Aviation 2019 Forum. American Institute of Aeronautics and Astronautics. DOI:10.2514/6.2019-3245.

34.

Halder

Benedict

. Nonlinear aeroelastic coupled trim analysis of a twin cyclocopter in forward flight. AIAA J 2021; 59: 305–319.

35.

Shrestha

. Experimental Investigation of a MAV-Scale Cyclocopter. PhD Thesis, University of Maryland, 2018. DOI:10.13016/S6EN-WSUK.

36.

Runco

Coleman

Benedict

. Development of the world’s smallest cyclocopter. In Proceedings of the AHS International’s 72nd Annual Forum Technology Display. West Palm Beach, Florida.

37.

Runco

Coleman

Benedict

. Design and development of a 30 g cyclocopter. J Am Helicopter Soc 2019; 64: 1–10.

38.

Runco

Benedict

. Flight dynamics model identification of a meso-scale twin-cyclocopter in hover. In Proceedings of the VFS International’s 77th Annual Forum Technology Display. Palm Beach, Florida, USA: Vertical Flight Society.

39.

Benedict

Jarugumilli

Lakshminarayan

et al. Effect of flow curvature on forward flight performance of a micro-air-vehicle-scale cycloidal-rotor. AIAA J 2014; 52: 1159–1169.

40.

Benedict

Mullins

Hrishikeshavan

et al. Development of a quad cycloidal-rotor unmanned aerial vehicle. J Am Helicopter Soc 2016; 61: 1–12.

41.

Hwang

Min

Lee

et al. Development of a four-rotor cyclocopter. J Aircr 2008; 45: 2151–2157.

42.

Ramsey

Benedict

. Design, development, and flight testing of a 25-kilogram quad-cyclocopter. In Proceedings of the VFS International’s 78th Annual Forum Technology Display. Fort Worth, Texas, USA: Vertical Flight Society.

43.

GmbH

. First free flight. https://www.youtube.com/watch?v=I2qkhmz4z1E, 2021. https://www.youtube.com/watch?v=I2qkhmz4z1E.

44.

Benedict

Chopra

. Design and development of an unconventional VTOL micro air vehicle: The cyclocopter. In George T, Islam MS and Dutta A (eds.) Micro- and Nanotechnology Sensors, Systems, and Applications IV. SPIE. DOI:10.1117/12.919482.

45.

Benedict

Gupta

Chopra

. Design, development, and open-loop flight-testing of a twin-rotor cyclocopter micro air vehicle. J Am Helicopter Soc 2013; 58: 1–10.

46.

Hrishikeshavan

Chopra

. Refined lightweight inertial navigation system for micro air vehicle applications. Int J Micro Air Vehicles 2017; 9: 124–135.

47.

Hrishikeshavan

Benedict

Chopra

. Identification of flight dynamics of a cylcocopter micro air vehicle in hover. J Aircr 2015; 52: 116–129.

48.

Georgy

Noureldin

Korenberg

et al. Modeling the stochastic drift of a MEMS-based gyroscope in gyro/odometer/GPS integrated navigation. IEEE trans Intell Transp Syst 2010; 11: 856–872.

49.

Thong

Woolfson

Crowe

et al. Dependence of inertial measurements of distance on accelerometer noise. Meas Sci Technol 2002; 13: 1163–1172.

50.

Runco

Benedict

. 60-gram quad cyclocopter (texas a&m university). https://www.youtube.com/watch?v=eL4gXdwA8vg, 2021.