Flight Testing A Micro Air Vehicle Using Morphing For ... - CiteSeerX

Flight Testing A Micro Air Vehicle Using Morphing For Aeroservoelastic Control




Mujahid Abdulrahim, Helen Garcia, Gregory F. Ivey, and Rick Lind



Micro air vehicles are able to fly in environments with little maneuvering room; however, such flight requires high agility and precision maneuvering. A current class of micro air vehicle is constructed with only rudder and elevator for control authority. Such surfaces are not optimal for flight control but ailerons can not be installed on the membrane used for wings on these vehicles. Morphing is used as an aeroservoelastic effector for control. Vehicles are constructed using threads to command a curl of the wings and using torque rods to command a twist of the wings. In each case, flight tests demonstrate the actuation causes sufficient deformation of the wing to result in significant control authority. The flight dynamics show turns and spins can be repeatedly performed using this aeroservoelastic control.

I. Introduction Micro air vehicles (MAV) are rapidly gaining attention in the flight test community. The small size and light weight of these vehicles make them especially attractive for some missions. The vehicles are quite portable so may be easily transported to remote locations. A MAV is inherently stealthy because of its size and quiet because of its electric propulsion. Furthermore, many types of sensors are being miniaturized and would fit within the payload capabilities of a MAV. Some areas of operation, such as urban environments, require a vehicle that is small but also extremely agile. Agility is obviously needed so the vehicle can maneuver within small areas and between obstacles in a dense environment. The vehicle must also be highly responsive to reject disturbances like wind gusts which may be considerable near buildings. The concept of aeroservoelastic tailoring through morphing is being studied for control on many vehicles. The basic concept of this approach involves actively changing the shape of a vehicle to affect the aerodynamics and, consequently, the flight dynamics. Such control inherently uses the control and actuation through the structural dynamics to affect the aerodynamics as an aeroservoelastic system. The resulting system has difficult dynamics so design and analysis of aeroservoelastic control is a challenge.1 Control through aeroservoelasticity, while not common, is certainly not a new concept. 5 The most famous example is arguably the wing twisting used by the Wright brothers to operate their flier. Academic studies have mainly focused on flutter control but several examples of maneuvering design have included the use of dynamic inversion 2 and robustness3 to account for the aeroservoelastic dynamics. In practice, the design is often limited to simply formulating notch filters to avoid instabilities but retain low-frequency performance. 4 The Active Aeroelastic Wing project has probably been the first aircraft to directly require design of an aeroservoelastic controller. 15 Morphing has more recently been considered a viable method of control. 6 The approach to change the shape of a vehicle will clearly affect the flight dynamics and static properties such as lift and drag. Issues remain to be studied 

 Graduate Student, [email protected], Student Member AIAA

 Graduate Student, [email protected], Student Member AIAA  Undergraduate Student, [email protected], Student Member AIAA

Assistant Professor, Department of Mechanical and Aerospace Engineering, 231 MAE-A Building, Gainesville FL, 32611,[email protected], Senior Member AIAA, University of Florida

1 of 17 American Institute of Aeronautics and Astronautics

as to the use of morphing for dynamic control. For instance, studies need to be performed ascertaining if a vehicle be morphed with sufficient magnitude and bandwidth to control the short period motion. In this case, the use of morphing as aeroservoelastic control for agility is the issue. This paper considers the use of morphing as an aeroservoelastic effector for control of a micro air vehicle. Several types of morphing are actually considered on several types of vehicles. A vehicle with 12 in wingspan has threads that curl the wings. Another vehicle with 24 in wingspan has torque rods that twist the wings. In each case, the morphing is shown to generate significant control authority for maneuvering. Properties of turns and spins are highlighted to demonstrate the effect of this aeroservoelastic control.

II. Morphing The concept of morphing is a fairly broad-ranging idea. A morphing aircraft is generally accepted to be an aircraft whose shape changes during flight to optimize performance. 6 Such changes might include span, chord, camber, area, thickness, aspect ratio, planform and any other metric related to shape. The morphing can even be applied to a control surface to eliminate hinges.29 The continuing maturation of materials and controls technology is leading to consideration of more extreme types of morphing for control. Aircraft can be morphed in several biologically-inspired ways that are appropriate for control. For instance, birds have very complex wing shapes during flight with large variations in sweep and twist along the entire span. Furthermore, the span itself changes based on flight condition for many birds. The complicated morphing is actuated using joints, such as elbow and wrist, along with the feathers to achieve many wing shapes. Programs by DARPA and NASA have been extensively studying the concept of morphing. These studies have shown the aerodynamic benefits; however, the use of morphing for control design has not been studied extensively. The limited research into control has actually considered static issues. An aircraft with morphing has been shown to have good performance when fixed at different configurations. 21 Control was considered for morphing in a related study but only considered the amount of actuation energy needed to cause morphing. 26 The amount of control authority was also studied for smart materials embedded in a structure in the context of steady-state open-loop controlled flight. 23 The issue of control design for morphing systems has partly been overlooked because of the lack of experimental testbeds. Many mechanisms for morphing have been constructed but they are not yet implemented on a flight vehicle. NASA has constructed a wing with morphing camber.24 A set of smart spars has been built that provide many types of morphing.19 Mechanisms have also been built such as a telescoping spar to morph aspect ratio 20 and an inflatable actuator to change sweep.28 In each case, the mechanisms are too heavy for a flight system so only wind tunnel testing of a static mount has been performed. An inflatable concept has actually been taken to flight 22 but the morphing is a single one-time inflation so is not used for maneuvering control. A project particularly relevant to the study of morphing for control is the F/A-18 Active Aeroelastic Wing (AAW). 15 The AAW is using wing twist to alter the aeroservoelastic dynamics and generate roll moments. The morphing is actually a passive effect resulting from moving leading-edge surfaces of the flexible wing. That testbed used a simple mechanism to generate morphing in order to study the resulting loads and control issues. A MAV is an especially appropriate platform for consideration of morphing. The flexibility of the membrane enables the shape of the wings to be easily changed. The resulting change will obviously affect the dynamics and act as an effector to provide control authority. As such, the morphing of a MAV adds very little cost but provides tremendous benefit. This paper will consider simple strategies to provide morphing to emphasize control issues. Essentially, the methods of morphing are not considered as much as the dynamics of morphing.

III. Vehicle Design Small, remotely piloted vehicles are adapted for morphing through simple modification of the structure. The vehicles are between 10 in and 24 in in wingspan and use a variety of structural members to both withstand flight loads and accomplish morphing deformation. Vehicles of this size are especially well suited as testbeds for simple morphing strategies because of the high strength of the actuators compared to flight loads. Lightweight MAV airframes

2 of 17 American Institute of Aeronautics and Astronautics

are subjected to relatively small air loads during flight and maneuvering. The small aerodynamic loads permit flexible materials to be used throughout the structure. Such flexible structures are easily deformed using existing actuators. In turn, simple morphing strategies for small air vehicles are readily developed and flight tested. A. Wing Curling Aircraft Morphing control was first implemented on a 10 in wingspan micro air vehicle. This vehicle used elevator and rudder surfaces for control, both of which were mounted on a conventional empennage. A folding, flexible wing was attached to the fuselage of each aircraft. The wing was mounted on cabane struts for the 10 in wingspan vehicle, and on mid-fuselage rods on the larger 12 in vehicle. The different wing position allowed each vehicle to undergo a slightly different type of morphing deformation. The two vehicles, shown in Figure 1, are essentially similar, apart from the wing position and the inclusion of rudder control on the 10 in vehicle. Otherwise, both vehicles accomplish morphing by curling the composite wings and changing the angle of incidence on one wing at a time. Turn control is accomplished by morphing a single wing to increase the aerodynamic forces and generate a roll moment in the direction opposite to the morphed wing. For instance. curling the left wing results in a positive (right-handed) roll rate and an ensuing right turn.

Figure 1. Wing curling morphing MAVs - 10 in wingspan high-wing aircraft (left) and 12 in span mid-wing aircraft (right)

The folding, flexible wing uses a strip of carbon fiber weave as the primary structural member. The strip extends approximately 1 in back from the leading edge of the wing, reaching the point of maximum camber. Thin strips of unidirectional carbon fiber are embedded within the leading edge strip and extend to the trailing edge. The entire structure is cured against a female mold to form the airfoil shape. Finally, an extensible latex membrane is adhered to the leading edge and battens. The result is a thin, undercambered wing surface having both flexibility in the battens and curling capability in the leading edge strip. The curved leading edge section is able to resist considerable bending loads in the positive direction. Under negative loading, however, the structure readily buckles and deforms in both curling and twisting. This type of wing structure is normally used for collapsing the wing of a MAV to fit into a small container. It is also well suited for morphing, given the compliance of the structure. The morphing is controlled by using small rotary actuators connected to hard points on the wing structure by tensioned Kevlar cables (Figure 2). As the actuator adjusts the tension on the cable, the wing deforms into a twisted form that is appropriate for flight control. Namely, the resulting shape increases the angle of incidence of the morphed wing and increases the lifting force produced. When one wing side is morphed, a lift differential is created, causing the aircraft to incur a roll rate. The extent and shape of the morphing can be adjusted by varying the amount of tension in the Kevlar lines or adjusting the location of the hardpoint on the wing. The shape is also dependent on the direction of the tensile force from the Kevlar, which is determined by the position of the actuator arm with respect to the wing hardpoint. A large vertical separation between these two points, as on the 10 in aircraft, produces a largely twisting motion. However, as the tensile force is applied in a more spanwise direction, as on the lower wing 12 in aircraft, the wing exhibits a predominantly curled motion during actuation (Figure 3). 3 of 17 American Institute of Aeronautics and Astronautics

Figure 2. Underside view of folding, flexible wing with tensioned Kevlar cables

Figure 3. Rear view of 12” MAV showing undeflected (left) and morphed wing (right)

Without twisting (and strictly curling), the wing morphing does not affect the roll trim to sufficiently control the aircraft. The 12 in airplane requires an additional Kevlar strand connected to a hardpoint on the trailing edge to produce a sufficient amount of wing twisting. Essentially, a single actuator output arm is used to tension two cables connected to two points on the wing. The resulting morphing produces a shape deformation that is suitable for lateral-directional control. The shape and thickness of the carbon fiber leading edge strip has a considerable effect on the morphing shape. The wings are somewhat sensitive to design or fabrication changes, where a disparity in the structure can cause a drastically different morphing shape. B. Wing Twisting Aircraft The concept of taking an existing design and retrofitting a morphing mechanism was extended to a 24 in vehicle of similar geometry (Figure 4). As with the smaller aircraft, the vehicle was based on conventional configuration and used an elevator and rudder for control. The primary difference between the two aircraft is the wing structure, which uses a nominally rigid unidirectional carbon fiber leading edge instead of the more compliant carbon weave. Additionally, the wing surface is a less extensible nylon film. Both the wing surface and leading edge structure reduce the magnitude of the deformation that results from flight loads. However, the wing retained the flexibility in the battens, which allows for wing warping via an aluminum torque-rod.

4 of 17 American Institute of Aeronautics and Astronautics

Property Wing Span Wing Area Wing Loading Aspect Ratio Powerplant Total Weight

Wing Twisting MAV 24 100   20.32   5.76 Brushless motor - 4.75 prop 400 

Wing Curling MAV 12 44   14.19   3.27 DC motor - 3.5 prop 123

Table 1. Properties of the 24 in and 12 in MAVs

Figure 4. 24” span wing twisting MAV

The torque-rod is affixed to a batten at approximately the 66% span position. Actuating this rod with a servo forces the wing to undergo a twisting deformation. Although the actuating point is localized to a single wing batten, the wing surface distributes the deformation over the entire wing. The magnitude of the twist deformation is largest at the actuation point and is tapered towards the wing tip and wing root. The majority of the deformation occurs across the wing battens and wing surface, although the leading edge also complies slightly during large morphing commands.

Figure 5. Underside view of 24” MAV wing showing aluminum torque rod

5 of 17 American Institute of Aeronautics and Astronautics

The primary difference in concept between the 24 in wing warping and the 12 in wing curling is the bidirectional, positive control of the shape deformation. The torque-rods are able to actuate the wing in twist in both trailing-edge up and trailing-edge down directions. Such control is not possible using the tensioned cables, as the mechanism relies on the wing structure to progressively buckle and curl. Should the wing leading edge deform too soon or too far, the cable would develop slack and deteriorate control of the wing shape. Since the wing warping aircraft uses a stiffer leading edge material, the problem of premature buckling or excessive deformation under aerodynamic load is largely eliminated. Thus, the control of the wing shape is largely a function of the actuator position, although the wing remaines free to deform slightly in response to airloads.

IV. Flight Testing and Handling Qualities A. Turns and Basic Controllability 1. Wing Curling The controllability of the wing curling aircraft is improved over a rudder-elevator equipped aircraft. The wing morphing facilitates flight path tracking over the majority of the flight envelope. At cruise airspeed, the morphing control is sufficient for small, stabilizing adjustments about trim in addition to large commands. Although some roll-yaw coupling results from the morphing deflection, the response remains greatly improved over rudder control for turns. The wing curling morphing exhibits good control response near the neutral, trim position. Small inputs are necessary in performing turns and in making slight adjustments to the flight path. Under these circumstances, the morphing provides an adequate level of control. Although the physical deformation of the wing surface is not necessarily linear, the aircraft responds predictably to various magnitudes of control input. In particular, the morphing is suitable for both commanding turns and for correcting for attitude perturbations from wind gusts or other disturbances. Roll controllability remains satisfactory throughout the airspeed range encountered during cruise, high-speed dives, and landing/approach phases. However, roll handling qualities tends to be quite sensitive to the location of the hardpoint on the wing and to the tension in the cable. Slight asymmetries in the right and left side cable tensions often contributes to difficulties in control and non-zero trim condition. Over a series of flights, the control response changed slightly from variations in the control linkages. Additionally, deterioration of the latex membrane noticeably reduced the wing surface tension. The natural rubber used in the latex material decayed when exposed to the sun. As a result, the reduced tension prevented the deformation from propagating smoothly throughout the wing structure. In turn, the twist deformation caused by the buckling remained localized around the hardpoint and reduced control effectiveness. 2. Wing Twisting Turn performance is considerably improved with the wing twist morphing on the 24 in aircraft. The torque-rod mechanism facilitates basic tasks such as commanding a bank angle and correcting for attitude perturbations. The aircraft response to small command inputs is more immediate and consistent than the wing curling aircraft. This improvement occurred in part due to the improved control power of the actuator over the morphing deflection. Additionally, the wing twist generates an anti-symmetric command that morphs the wing in both positive and negative direction on both wings simultaneously. In addition to substantially improving control power, the anti-symmetric morphing reduces yaw coupling compared to the wing curling morphing. Minimal rudder corrections are required to maintain turn coordination during maneuvering. The improved morphing mechanism of the 24 in aircraft contributed to vastly improved handling qualities. In particular, the aircraft performance in gusty wind conditions benefits from the positive control over the morphing. High frequency control tasks such as attitude correction and stabilizing control are easily performed.

6 of 17 American Institute of Aeronautics and Astronautics

B. 360  Rolls Continuous 360  roll maneuvers are performed for each aircraft to determine the maximum roll rate achievable with morphing. The tests help determine the response of the aircraft to a long morphing command. They are also useful in identifying any obvious cross-coupling in pitch and yaw that occur during the maneuver. Roll tests are performed with the aircraft at trimmed, straight and level flight at cruise airspeed. Full morphing is commanded to either right or left directions and held until the aircraft has completed a 360  rotation along the roll axis. 1. Wing Curling The wing curling morphing provides a sufficient level of control to effect a complete roll. Despite the asymmetry, the aircraft experiences a relatively small divergence in flight path during the maneuver. Under maximum morphing deflection, a stabilized roll rate is achieved within 0.5 seconds. Maximum roll rate is approximately  , achievable in both left and right roll directions. During the roll maneuver, the aircraft prescribes an almost helical flight path. A slight gain in altitude is incurred halfway through the maneuver, followed by a descending return to level flight. At the completion of the roll, the aircraft is typically pitched down slightly and may have a small heading change. 2. Wing Twisting The aircraft is able to achieve a high level of roll performance using wing twist morphing. The performance is improved compared to wing curling, rudder, or even conventional ailerons. At the maximum morphing deflection of approximately 10  twist, the aircraft achieves a stabilized roll rate of !" within 0.25 seconds. At nearly 3 rolls per second, aircraft experiences very little flight path divergence or cross-coupling. During the maneuver, the aircraft wings become visibly blurred from the rotation, while the fuselage appears to rotate about the centerline axis. Similar MAVs using rigid wings and hinged conventional ailerons were able to achieve a maximum measured roll rate of #%$  . Recovery from the roll maneuver is also within 0.25 seconds of neutralizing morphing command. Roll performance changes slightly over the airspeed range, with higher airspeeds exhibiting higher roll rates. However, the control remains positive and sufficient from near-stall speed up to the maximum dive speed. C. Spins Control of a morphing vehicle beyond the stall boundaries is another relevant facet of the flight dynamics. A greater degree of control over the vehicle geometry may improve stall/spin avoidance or, conversely, even command a developed spin maneuver. The large degree of control afforded by the morphing mechanism could be beneficial in generating anti-spin forces and recovering from a stabilized spin. The spin characteristics of the vehicles are investigated by manually-piloted maneuvers. Spin modes are identified using classical spin entry techniques.18 The individual modes are produced by trial-and-error and have been reproduced over several flight tests. The following discussion presents some of the predominant spin types encountered during these flight tests. Note these spins all use morphing because the spin characteristics were not nearly as significant using only rudder and elevator. 1. Wing Curling Spin modes of the 10 in and 12 in vehicles were initially encountered during flight tests of aggressive maneuvers. Large morphing commands coupled with positive elevator commands resulted in sudden and violent rolls opposite to the commanded direction. This behavior was most often observed at lower airspeeds, leading to the identification of a considerable deficiency in the wing curling method. The wing curling morphing is able to control roll rate primarily by increasing the angle of attack of the morphed wing. The deformation also encompasses spanwise curling, although the effect of twisting to increase the incidence

7 of 17 American Institute of Aeronautics and Astronautics

appears to be the predominant factor controlling roll moment. Under typical flying conditions and moderate morphing deflections, the morphing is able to generate a proverse lift differential (i.e. in the commanded direction) and effect a roll rate change. This is the same characteristic described in the turn and roll performance sections. However, as the morphing command becomes large (causing a large change in incidence angle) and is coupled with positive elevator deflection, the angle of attack of the morphed wing can exceed the stall angle, causing a severe loss of lift in the deformed region. The resulting lift differential between the two wings becomes adverse to the commanded direction, leading to an opposite-direction roll. While this is occasionally recoverable, such an asymmetric stall typically diverges into a spin. The overwhelming majority of these encountered spins were terminal, resulting in the aircraft impacting the ground in a stabilized spin. Recovery from a stall-spin is quite difficult for a number of reasons. The researchers suspect that the stall resulting from excessive morphing position causes a collapse of the low-pressure region above the wing. This collapse reduces the tension on the control cable and prevents the wing from properly extending and returning to trim flight condition. Additionally, the morphing mechanism permits the pilot to strictly increase the angle of attack of the wings. Thus, in opposing the roll rate incurred during a spin, the pilot must increase the angle of attack of the opposite (unstalled) wing, which may further decay the controllability of the aircraft. Some spins tests resulted in a successful recovery to level flight. These spins were typically initiated using a gentler stall entry and smaller magnitudes of elevator and morphing command inputs. Flight path during the spins resembled a descending helical spiral. Recovery procedure entailed neutralizing control surface and morphing deflection and allowing the MAV to self-recover from the spin. Once stabilized, elevator command was used to pull up from a vertical dive to level flight. However, because of the difficulties in consistently recovering from a departure, spins were not considered as useful maneuvers for this type of MAV. 2. Wing Twisting Figure 6 shows the command and rotation rates during a conventional spin. This maneuver is initiated from level flight by commanding positive elevator to increase the pitch rate and angle of attack. Right rudder command is then applied to generate a yawing moment as the aircraft approaches stall. In this case, the yaw causes an asymmetric stall and starts the spin rotation. The aircraft response is relatively constant throughout the maneuver, although the roll rate tends to build up as the flight path changes from level to a vertical. The autorotation continues as long as the positive elevator and rudder commands are held. Once the commands are neutralized, the rotation slows and comes to a stop with little or no opposite rudder input. Positive elevator is used to recover the aircraft to level flight at 363 seconds. Although this type of spin has been experienced several times, the entry procedures tend to be difficult to reproduce. Specifically, applying rudder command at a low angle of attack (too early) prevents a stall from developing and results in a high-speed spiral dive. Both wind tunnel and CFD analysis have shown that the thin-undercambered airfoils used on the vehicle have delayed stall response. This affords such vehicles increased resistance to stall-spin departure, at least for positive loadings. The effect of morphing on positive (upright) spins is to accelerate the onset of the spin and to assist in the recovery process. This effect is most pronounced during cross-coupled controls, where the rudder direction is opposite to that of the morphing. In such a case, the high angle of attack at the inside wing tip is further increased by the morphing actuation, leading to an observed stall-spin. Releasing the morphing command effectively reduces the wing angle of attack and produces nearly immediate recovery from an upright, conventional spin. Conventional spins are also performed with negative (down) elevator actuation to produce a starkly different response. In particular, the spin modes observed are of considerably higher energy. The rotation rates of a negative spin compared with an upright spin tend to be between 2 to 6 times greater. Based on rudimentary analysis, the stall characteristics of a thin under-cambered wing at negative angles of attack are far more severe than the characteristics at high angles of attack. In flight, the airplane is observed to have a very immediate and violent response to large negative elevator commands. Such an input is believed to cause a negative stall quickly, where any asymmetry about the yaw axis produces a large rate of rotation. Figure 7 shows an identified negative spin mode initiated by a morphing command with elevator and rudder. At 401 seconds, the aircraft responds to the constant control deflection by building up rotation rates on all three axis. The

8 of 17 American Institute of Aeronautics and Astronautics

40

50

20

Response (deg/s)

Command (deg)

30

10 0

−10 −20 elevator rudder morphing

−30 −40 359

360

361

362

Time (s)

363

0

−50 roll rate pitch rate yaw rate

−100 359

364

360

361

362

Time (s)

363

364

Figure 6. Pilot Commands (left) and Responses (right) during conventional spin

entry into the maneuver is relatively gradual and only after one second of control inputs have the pitch, roll, and yaw rates become significant. This particular type of spin tends to stabilize independently of the initial pro-spin control deflections. At t=402 seconds, the controls are released, while the aircraft continues to spin. The application of positive elevator (for recovery) shortly afterwards appears to maintain the spin for some time. It is only with corrective opposite rudder command that the aircraft arrests the rotation and recovers from the spin. It is difficult to draw solid conclusions from this spin sequence. However, the researchers attribute the two distinct modes observed to be a case of primary and secondary spin characteristics, where the latter is caused by a premature recovery attempt. Similar spins have been observed from in both left and right directions. 40

elevator rudder morphing

50

20

Response (deg/s)

Command (deg)

30

10 0

−10 −20

−50 roll rate pitch rate yaw rate

−100

−30 −40 400

0

401

402

403

Time (s)

404

400

405

401

402

403

Time (s)

404

405

Figure 7. Pilot Commands (left) and Responses (right) during spin

Alternatively, Figure 8 shows a considerably different spin behavior. Although initiated by commands similar to the previous spins, this type of spin exhibits a cyclic or periodic motion. It is perhaps with the timing of the control inputs that a difference can be found. Whereas in Figure 9, the elevator input lagged behind the rudder and morphing inputs, the spin depicted by Figure 8 shows the elevator leading slightly. The precise effect this has on the airflow is unknown. However, the resulting aircraft response is shown to be 6 times greater in magnitude than a conventional spin. From level, trimmed flight, the aircraft is subjected to full left wing morphing, full left rudder, and full negative elevator command. The initial reaction of the aircraft is to pitch down at a constant rate and incur a left roll and 9 of 17 American Institute of Aeronautics and Astronautics

yaw from the wing morphing and rudder deflections. Once the wing has reached the negative stall angle, presumably facilitated by the deflected wing, a rapid spin ensues, nearly doubling the roll and yaw rates and reducing pitch rate. This pattern is repeated four times throughout the spin, all while pilot commands are held constant. Each cycle is proceeded by a period of low momentum, followed by a sharp change in pitch rate along with peaks in both the roll and yaw rates. While the dynamics of such a maneuver are not very well understood, it appears that the morphing of the wing plays a large roll in both inducing and recovering from the spin. For instance, similar spin entries performed without morphing are characterized by considerably lower rotation rates and a continuation of the spin after command inputs are neutralized. However, the recovery of this cyclic spin mode occurs nearly immediately after the controls are neutralized. As seen at t = 176 in Figure 8, the aircraft is at the period of highest moment during return to neutral command. The rotation rates continue to follow the characteristic spike pattern and finally converges to zero rotation rates. 40

elevator rudder morphing

50

20

Response (deg/s)

Command (deg)

30

10 0

−10 −20

0

−50

−100

−30 −40 171

roll rate pitch rate yaw rate

172

173

174

Time (s)

175

176

171

177

172

173

174

Time (s)

175

176

177

Figure 8. Pilot Commands (left) and Responses (right) during cyclic spin

In flight, this has the effect of stopping the aircraft in mid-rotation. Unlike the other spin modes observed, the cyclic spin mode has no apparent recovery apart from neutralizing the controls. The aircraft will continue to the end of a given cycle, cease rotation, and simply fly away. The nose-down recovery typical of other spin modes is contrasted with an immediate recovery to level flight. The usefulness of the cyclic spin mode depicted in Fig. 8 is perhaps questionable, although it may give rise to a different mode of maneuvering for morphing aircraft. For instance, the above maneuver may be useful for a controlled vertical displacement. On initiating the entry, the airspeed quickly decays and starts the aircraft on a relatively slow vertical flight path. During this portion of the maneuver, the aircraft incurs a series of high rate of rotations, each separated by a period of low momentum. As evidenced by the recovery from the maneuver, this period can be used to recover the aircraft into stable flight. While previous spin modes required corrective rudder and significant altitude losses for recovery, this cyclic spin mode stopped once the controls were neutralized. Attitude and airspeed entry conditions into the spin trials have been observed to have some impact on the stabilized spin modes; however, accurate measurements of the entry conditions were not possible. The lack of pressure sensors on the airframe precluded the gathering of such data. Excitation of a particular spin mode depended on the pilot ability to position the aircraft properly based on control feel and vehicle observations. The spin entry maneuvers were also attempted for other control combinations. Specifically, cyclic spins were attempted without wing twisting by using negative elevator and rudder deflection. These trials resulted in a stabilized spin but with considerably lower rotation rates than the cyclic spin. Additionally, this mode did not exhibit the periodic behavior achieved through wing twisting during a spin.

10 of 17 American Institute of Aeronautics and Astronautics

V. Modeling A. Wing Curling Flight data from the vehicle is analyzed to estimate models of the flight dynamics. Several techniques were attempted to estimate these models, including system identification27 and parameter estimation,32 but with limited success. This vehicle is particularly difficult to model because the morphing causes time-varying asymmetries which violate many assumptions used by standard routines. Furthermore, the estimation is difficult because of limited flight data. The MAV is equipped with only gyros and accelerometers but the flight data from the accelerometers is actually too noisy to be useful for modeling. Thus, several critical measurements, such as angle of attack and angle of sideslip, are not available. Some dynamics are not easily observable, especially in the presence of noise, using only the available sensors. A nonlinear auto-regressive model is used to represent the flight dynamics. The general form of this model is shown in Equation 1. This model relates the gyro measures of roll rate, &('*),+ , pitch rate, -'.),+ , and roll rate, /'*),+ , to the morphing command, 0213'.),+ , and elevator command, 0545'.),+ , at the sampling instance of ) . The matrices, 687:9=?= and @ 7 9A;>=5?  , represent the dynamics. BC

JLK &('*)FEG!5+ D

M

-'.)FEH!2+ /'.)IEH!2+ BC

JLK

BC

JLK

N

&('.)+ 6PO D

-'.),+

M

EQ6 

/'.),+ BC

JLK

&('.)+^-'.)+ EP6]\ D

-'.),+^/'.),+

M

D

-'.)TRS!2+

JLK

0c13'.),+ 0"1T'*)TR[!5+ed

&('*),+ M

EU6 =

BC

EQ6]` D

DWV -'*),+ V -'.),+

-'.)3RS!2+^/'.)3RS!2+

JLK

&('*),+ M

&Z'.)IR[!5+ EQ6YX

&Z'.)TRS!2+

DWV -'*)TR[!5+ V -'*)TR[!5+

V /'*),+ V /'*),+ /'.)3RS!2+ BC JLK V V &('*)IRS!2+a-'.)3RS!2+

/'*),+_&Z'.),+ EP@8O3b

BC

&('*)IRS!2+

M

V /'*)TR[!5+ V /'*)3R[!5+ V

V

M

/'.)3RS!2+_&Z'.)3RS!2+ EU@ 

0c13'.)+ b

0"1T'*),+

V

0"1T'*V )TR[!5+ V 0"1T'*)TR[!5+fd

EU@ =

V

b

024'.)+ 0"4'.)3RS!2+fd

(1)

The model in Equation 1 contains quadratic terms of the rates and commands. Such quadratic terms are included to account for unknown relationships between the wing shape and the aerodynamics. In this case, the terms utilize an absolute value to allow the contributions from the quadratics to change in sign. The model in Equation 1 also contains coupling terms. These terms multiply the gyro measurements by each other. The standard equations of motion for a rigid-body aircraft include coupling terms which scale by the moments of inertia.33 This MAV is obviously asymmetric during the morphing so the coupling are essential. Finally, Equation 1 computes the update to the gyro measurements as a function of the measurements from two previous sampling times. These terms are included to account for the time-varying nature of the dynamics which arise by altering the wing shape. The dynamics are assumed to be sufficiently described by two sampling times although a rigorous study of the sampling times was not conducted. The values of the matrices, 6 7 and @ 7 , in Equation 1 are determined by a least-squares fit to the flight data. The resulting model is used to simulate the responses to the morphing and elevator commands. Such responses are shown in Figure 9. The responses in Figure 9 demonstrate the model captures the basic trend of the dynamics but is not completely accurate. The predicted responses are not perfect matches to the measured responses but yet they clearly show similarities. Thus, the model indicates the time-varying asymmetries associated with the morphing causes nonlinearities and coupling in the flight dynamics of this MAV.

11 of 17 American Institute of Aeronautics and Astronautics

Pitch Rate (deg/s)

Roll Rate (deg/s)

10

8

5

data sim

5 0 −5

data sim

0

−5

−10 −15 0

data sim

6

Yaw Rate (deg/s)

15

4 2 0 −2 −4 −6

1

2

3

Time (s)

4

5

−10 0

6

1

2

3

Time (s)

4

5

−8 0

6

1

2

3

4

Time (s)

5

6

Figure 9. Measured and Predicted Responses for Roll Rate (left), Pitch Rate (middle) and Yaw Rate (right)

B. Wing Twisting

150

150

10

100

100

5 0 −5 −10 −15 0

Yaw Rate (deg/sec)

15

Roll Rate (deg/sec)

Rudder Command

Flight testing of the active wing-shaping 24 in MAV is performed in the open area of a radio controlled (R/C) model field during which wind conditions range from calm to 7 knots throughout the flights. Once the flight control and instrumentation systems are powered and initialized, the MAV is hand-launched into the wind. This launch is an effective method to quickly and reliably allow the MAV to reach flying speed and begin a climb to altitude. This airplane is controlled by a pilot on the ground who maneuvers the airplane visually by operating an R/C transmitter. The data acquisition system begins recording as soon as the motor is powered. This aircraft design allows either rudder or wing shaping to be used as the primary lateral control for standard maneuvering. The airplane is controlled in this manner through turns, climbs, and level flight until a suitable altitude is reached. At altitude, the airplane is trimmed for straight and level flight. This trim establishes a neutral reference point for all the control surfaces and facilitates performing flight test maneuvers. Open-loop data is taken to indicate the flight characteristics of the MAV. Specifically, the rates and accelerations are measured in response to doublets commanded separately to the servos. Several sets of doublets are commanded ranging in magnitude and duration to obtain a rich set of flight data. The dynamics of the MAV in response to rudder commands is investigated to indicate the performance of the traditional configuration for this MAV. A representative doublet command and the resulting aircraft responses are shown in Figure 10.

50 0 −50 −100 −150

1

2

3

Time(sec)

4

5

6

−200 0

50 0 −50 −100 −150

1

2

3

Time(sec)

4

5

−200 0

1

2

3

Time(sec)

4

5

Figure 10. Doublet Command to Rudder (left), Roll Rate response (middle), and Yaw Rate response (right)

The roll rate and yaw rate measured in response to this command are shown in Figure 10. The roll rate is sufficiently large and indicates the rudder is able to provide lateral-directional authority; however, the yaw rate is clearly larger than desired. Actually, the yaw rate is similar in magnitude to the roll rate so the lateral-directional dynamics are very tightly coupled. The effect of the rudder in exciting the dutch roll dynamics is clearly evidenced in this response. 12 of 17 American Institute of Aeronautics and Astronautics

Doublets, such as the pulse sequence shown in Figure 11, are also commanded to the morphing servo.

Roll Rate (deg/sec)

Morphing Command

6 4 2 0 −2 −4

150

150

100

100

50 0 −50 −100

−6

−150

−8 0

−200 0

0.5

1

1.5

Time(sec)

2

Yaw Rate (deg/sec)

8

50 0 −50 −100 −150

0.5

1

Time(sec)

1.5

2

−200 0

0.5

1

Time(sec)

1.5

2

Figure 11. Doublet Command to Wing twist morphing (left), Roll rate response (middle), and Yaw Rate response (right)

150

150

100

100

Yaw Rate (deg/sec)

Roll Rate (deg/sec)

The roll rate and yaw rate in Figure 11 are measured in response to the doublet. These measurements indicate the roll rate is considerably higher than the yaw rate. Thus, the morphing is clearly an attractive approach for roll control because of the nearly-pure roll motion measured in response to morphing commands. The data from open-loop flights is then used to approximate a linear time-domain model using an ARX approximation.27 This model is generated by computing optimal coefficients to match properties observed in the data. The assumption of linearity is reasonable since the maneuvers are small doublets around a trim condition. The resulting model, having poles at -4.95 and -0.1194, is used to simulate responses of the aircraft. The simulated and measured values of roll and yaw rates are shown in Figure 12.

50 0 −50 −100 −150 −200 0

50 0 −50 −100 −150

0.5

1

Time(sec)

1.5

2

−200 0

0.5

1

Time(sec)

1.5

2

Figure 12. Simulated ( gfg>g ) and Actual (—) Roll Rate (left) and Yaw Rate (right) Responses to a Doublet

The simulated responses show good correlation with the actual data. The model is thus considered a reasonable representation of the aircraft. The existence of such a model is important for future design of autopilot controllers but it is also valuable for interpreting the morphing. Essentially, the ability to identify a linear model with poles relating to the roll convergence and spiral convergence modes indicate the aircraft with morphing acts like an aircraft with ailerons.

13 of 17 American Institute of Aeronautics and Astronautics

VI. New Morphing Aircraft Designs A. Multi-point Wing Shaping Based on the success of the wing warping aircraft, an additional vehicle was developed to incorporate increased control of the wing shape. Specifically, the vehicle employed a series of concentric torque-tube spars to control the wing incidence angle at four points along the span. Such a mechanism extended the idea of wing twisting beyond a single actuation point to a morphing of the entire wing surface. Figure 13 shows the wing undergoing morphing to both the wingtip spar tube and to both wingtip and midboard spar tubes. The deformation is visually apparent by examining the light reflections off of the leading edge and the shape of the trailing edge.

Figure 13. Wing shaping MAV showing neutral position (left), wingtip morphing (middle), and full wing morphing (right)

Concentric tube spars act as both primary load-bearing members and as control linkages (torque-tubes). A large diameter tube is fixed to the fuselage and acts as a bearing support for the rotating spars. The root section of the wing surface is also attached to this tube, creating an immobile joint between the inboard wing surface and fuselage. Two smaller tubes, one within the other, are supported by the fixed tube. The smallest tube extends the full span, while the center tube extends to the 60% position. Each of the outboard and midboard spars is actuated in twist via servos mounted in the fuselage (Figure 14). Each servo is then able to command the incidence angle of the corresponding wing section independently. A flexible wing surface is attached to each of the three wing spar tubes. Attachment points near the spar joints are left unconstrained in pitch angle. This freedom allows the incidence to smoothly taper between the rigidly attached sections of the wing surface. This structure permits twist morphing of each controlled wing section from Rh!" to ET!"% incidence angle. Each of the four wing sections are commanded independently, allowing for considerable differential or collective configurability.

Figure 14. Spar torque-tube morphing actuators. The 4 front servos rotate concentric spar sections, aft 2 control rudder and elevator

14 of 17 American Institute of Aeronautics and Astronautics

The aircraft has undergone basic performance and handling flight tests. Initial results indicate that the aircraft is capable of achieving significantly higher performance levels than the previous morphing case studies. Adequate roll control is achieved by differentially actuating the wingtip spars. The handling qualities and maximum roll rate are similar to the 24 in wing twisting aircraft. Actuating the entire wing differentially (i.e. using both wingtip and midboard sections), achieves roll rates and performance measures considerably higher. As with the wing twisting aircraft, at maximum roll rate, the multi-point wing shaping vehicle becomes difficult to see and establish orientation. The morphing is also being considered for use in conjunction with other control surfaces. Basic flight tests of combining collective midboard wing deflection with elevator command have shown potential for improvement in pitch rate performance. Additionally, this morphing may be suited for quasi-statically reconfiguring the wing twist to optimize spanwise lift distribution in flight. Such techniques are currently used by sailplane and commercial jet pilots to alter the lift properties of the wing for cruise, steep descent, and maximum performance flight phases. B. Variable Gull-Wing Angle Preliminary flight tests have been completed of a biologically-inspired variation of the 24” morphing aircraft. This MAV incorporates a variable angle gull-wing mechanism, Figure 15, that is used to change the angle between the inboard and outboard wing sections in flight. The mechanism uses a linear lead-screw actuator to quasi-statically vary the gull-wing position. The quasi-static morphing is used to investigate the effect of such a deformation on the flight performance of the vehicle. Specifically, it is of interest for an aircraft with an expanded flight envelope, capable of performing low airspeed, precise maneuvering, as well as high speed dashes and long range endurance flights.

Figure 15. Variable Gull-Wing Angle MAV. Negative gull-wing (left), neutral gull-wing (middle), and positive gull-wing (right)

Flight test results have shown that the gull-wing angle considerably affects basic performance metrics of the aircraft such as glide ratio, stall characteristics, climb ratio, and handling qualities. C. Airigami - Folding wing/tail A quasi-static morphing has also been implemented on a tandem-wing micro air vehicle, Figure 16, to allow the aircraft to achieve two distinct mission requirements in a single flight. The aircraft is designed to achieve stable, controllable forward flight for climb, cruise, and loiter phases, then transition to reverse flight for a slow, vertical descent. A single control actuator is used to sweep both front and aft wings forward, in addition to collapsing and extending vertical stabilizer surfaces. The aircraft incorporates a dual-wing sweep angle morphing to change the location of the aircraft center. The wings are designed to sweep far enough forward such that the neutral point becomes forward of the center of gravity. In this configuration (Figure 17), forward flight is destabilized and reverse flight is stabilized. In order to improve reverse flight stability, the wing sweep incorporates a collapsing vertical stabilizer on the aft wing and an expanding stabilizer on the forward wing. Each stabilizer is initially built into the wing structure and allowed to along fold nylon hinges. Reverse flight is achieved only in descents with a near vertical flightpath. As such, the thrust from the propeller serves as both a drag producer and as a stabilizing device. The primary purpose of the wings and vertical stabilizer during this descent profile is to prevent the vehicle from diverging from the vertical attitude. In this orientation, the thrust serves to directly counteract the weight of the aircraft and slow the sink rate. The current powerplant uses a DC electric motor with a 4:1 gear reduction to turn a 4” plastic prop. The thrust to weight ratio of the aircraft is slightly less than one, allowing for a substantial reduction in the sink rate at full throttle. Alternative motor options may be 15 of 17 American Institute of Aeronautics and Astronautics

Figure 16. Top view (left) and side view (right) of Airigami showing unswept, forward flight configuration

Figure 17. Top view (left) and side view (right) of Airigami, showing swept and folded reverse flight configuration

used to increase thrust to weight ratio to greater than one. In such a case, the thrust could be used to achieve a zero sink rate and hover the aircraft during the descent phase. Although the aircraft is designed primarily for vertical reverse flights, other descent modes such as a controlled flat spin or high-alpha, oscillatory falling leaf mode may be possible with the sweep morphing.

VII. Conclusions This paper demonstrates that morphing is particularly suitable for a class of micro air vehicles. The membrane wings on these vehicles can be morphed with little power but with significant benefits. Mechanisms that are relatively simple are shown to cause large deformations using different strategies. In each case, the morphing provides aeroservoelastic control that is clearly adequate for maneuvering. Flight tests demonstrate the morphing is able to command turns and spins with sufficient authority for precision maneuvering. As such, the wing shaping is an enabling technology providing some level of mission capability to this class of MAV.

References j

i

M.J. Brenner, Aeroservoelastic Modeling and Validation of a Thrust-Vectoring F/A-18 Aircraft, NASA-TP-3647, September 1996. I.M. Gregory, “Dynamic Inversion to Control Large Flexible Transport Aircraft,” AIAA-98-4323, 1998.

16 of 17 American Institute of Aeronautics and Astronautics

k

S.M. Joshi and A.G. Kelkar, “Inner Loop Control of Supersonic Aircraft in the Presence of Aeroelastic Modes,” IEEE Transactions on Control Systems Technology, Vol. 6, No. 6, November 1998, pp. 730-739. l B.D. Caldwell, “FCS Design for Structural Coupling Stability,” The Aeronautical Journal, December 1996, pp. 507-519. m E. Livne, “Integrated Aeroservoelastic Optimization: Status and Direction,” Journal of Aircraft, Vol. 36, No. 1, January-February 1999, n pp. 122-145. R.W. Wlezien, G.C. Horner, A.R. McGowan, S.L. Padula, M.A. Scott, R.J. Silcox, and J.O. Simpson, “The Aircraft Morphing Program,” AIAA-98-1927, April 1998. o J.B. Davidson, P. Chwalowski, and B.S. Lazos, “Flight Dynamic Simulation Assessment of a Morphable Hyper-Elliptic Cambered Span Winged Configuration,” AIAA-2003-5301, August 2003. p P.G. Ifju, D.A. Jenkins, S.M. Ettinger, Y. Lian, W. Shyy, and M.R. Waszak, “Flexible-Wing Based Micro Air Vehicles,” AIAA-2002-0705, January 2002. q H. Garcia, M. Abdulrahim, and R. Lind, “Roll Control for a Micro Air Vehicle using Active Wing Morphing,” AIAA-2003-5347, August 2003.isr Y. Lian and W. Shyy, “Three-Dimensional Fluid-Structure Interactions of a Membrane Wing for Micro Air Vehicle Applications,” AIAA2003-1726, April 2003.. iti M.R. Waszak, L.N. Jenkins, and P.G. Ifju, “Stability and Control Properties of an Aeroelastic Fixed Wing Micro Air Vehicle,” AIAA-20014005,i.j August 2001. G.A. Fleming, S.M. Bartram, M.R. Waszak, and L.N. Jenkins, “Projection Moire Interferometry Measurements of Micro Air Vehicle Wings,” Proceedings of the SPIE International Symposium on Optical Science and Technology, Paper 448-16, August 2001. isk S.M. Ettinger, M.C. Nechyba, P.G. Ifju, and M.R. Waszak, “Vision-Guided Flight Stability and Control for Micro Air Vehicles,” Proceedings of thei l IEEE International Conference on Intelligent Robots and Systems, October 2002, pp. 2134-2140. M.R. Waszak, J.B. Davidson, and P.G. Ifju, “Simulation and Flight Control of an Aeroelastic Fixed Wing Micro Air Vehicle,” AIAA-20024875,i.m August 2002. E.W. Pendleton, D. Bessette, P.B. Field, G.D. Miller, and K.E. Griffin, “Active Aeroelastic Wing Flight Research Program: Technical n and Model Analytical Development,” Journal of Aircraft, Vol. 37, No. 4, 2000, pp. 554-561. Program i S. Tung, and S. Witherspoon, “EAP Actuators for Controlling Space Inflatable Structures,” AIAA-2003-1741, April 2003. i o M.J. Solter, L.G. Horta, and A.D. Panetta, “A Study of a Prototype Actuator Concept for Membrane Boundary Control,” AIAA-2003-1736, Aprilis2003. p R.W. Stone, and B.E. Hultz, Summary of Spin and Recovery Characteristics of 12 Models of Flying-Wing and Unconventional-Type Airplanes, NACA-RM-L50L29, March 1951. isq M. Amprikidis and J.E. Cooper, “Development of Smart Spars for Active Aeroelastic Structures,” AIAA-2003-1799, 2003. j r J. Blondeau, J. Richeson and D.J. Pines, “Design, Development and Testing of a Morphing Aspect Ratio Wing using an Inflatable Telescopic Spar,”jui AIAA-2003-1718. J. Bowman, B. Sanders and T. Weisshar, “Evaluating the Impact of Morphing Technologies on Aircraft Performance,” AIAA-2002-1631, 2002.jtj D. Cadogan, T. Smith, R. Lee and S. Scarborough, “Inflatable and Rigidizable Wing Components for Unmanned Aerial Vehicles,” AIAA2003-1801, 2003. j k C.E.S. Cesnik and E.L. Brown, “Active Warping Control of a Joined-Wing Airplane Configuration,” AIAA-2003-1716, 2003. j l J.B. Davidson, P. Chwalowski and B.S. Lazos, “Flight Dynamic Simulation Assessment of a Morphable Hyper-Elliptic Cambered Span Winged Configuration,” AIAA-2003-5301, 2002. jtm n J.M. Grasmeyer and M.T. Keennon, “Development of the Black Widow Micro Air Vehicle,” AIAA-2001-0127, 2001. j C.O. Johnston, D.A. Neal, L.D. Wiggins, H.H. Robertshaw, W.H. Mason and D.J. Inman, “A Model to Compare the Flight Control Energy Requirements of Morphing and Conventionally Actuated Wings,” AIAA-2003-1716, 2003. j o L. Ljung, System Identification, Prentice Hall, Englewood Cliffs, NJ, 1987. j p P. de Marmier and N. Wereley, “Morphing Wings of a Small Scale UAV Using Inflatable Actuators for Sweep Control,” AIAA-2003-1802. j q B. Sanders, F.E. Eastep and E. Forster, “Aerodynamic and Aeroelastic Characteristics of Wings with Conformal Control Surfaces for Morphingktr Aircraft,” Journal of Aircraft, Vol. 40, No. 1, January-February 2003, pp. 94-99. D. Viieru, Y. Lian, W. Shyy and P. Ifju, “Investigation of Tip Vortex on Aerodynamic Performance of a Micro Air Vehicle,” AIAA-20033597,kv2003. i M.R. Waszak, L.N. Jenkins and P. Ifju, “Stability and Control Properties of an Aeroelastic Fixed Wing Micro Aerial Vehicle,” AIAA-20014005.k^j K.W. Iliff, “Aircraft Parameter Estimation,” NASA-TM-88281, 1987. ktk R.C. Nelson Flight Stability and Automatic Control McGraw Hill, Boston, MA, 1998.

17 of 17 American Institute of Aeronautics and Astronautics