Shape Memory Polymer-Based Flexure Stiffness Control.pdf

Embed Size (px)

Citation preview

  • 8/14/2019 Shape Memory Polymer-Based Flexure Stiffness Control.pdf

    1/4

    IEEE TRANSACTIONS ON ROBOTICS, VOL. 28, NO. 4, AUGUST 2012 987

    Communication

    Shape Memory Polymer-Based Flexure Stiffness Control

    in a Miniature Flapping-Wing Robot

    Lindsey Hines, Veaceslav Arabagi, and Metin Sitti

    AbstractAn active flexural hinge has been developed and incorporatedinto the transmission of a prototype flapping-wing robot. The multilayeredflexure, which is constructed from a shape memory polymer and a poly-imide film, showed controllable stiffness under change in temperature. Atroom temperature,the flexure hada bending stiffnessof 572 mNmm;whenwarmed to 70 C, the stiffness was 11 mNmm. The resulting single-wingflapping system demonstrated up to an 80% change in generated lift with-out modification of thewaveformof themain driving piezoelectric actuator.Such active stiffness tunable flexure joints could be applied to any flexuralminiature mobile robot and device mechanisms.

    Index TermsFlapping flight, flexural hinge, shape memory polymer(SMP).

    I. INTRODUCTION

    With the recent emergence of miniature robots and portable mo-

    tion mechanisms down to centimeter or millimeter size, bending sheet

    joints, compact and frictionless flexures that act as rotational joints,

    have become indispensable and have been used in a range of applica-

    tions and size scales [1][3]. As flexure behavior is based on cantilever

    bending, the flexure can be manufactured with different geometry or

    with various materials to achieve compliance about intended loading

    axes. However, these flexures are typically made of passive polymer

    materials where the joint stiffness does not change in the small deflec-

    tion regime. Where weight or size constrains a system, active flexural

    hinges can allow control without additional traditional driving actua-

    tors. In miniature flapping-wing-based flying robots, where research

    continues on insect-scale systems [4][7], they become especially use-

    ful. If multiple wings are driven with a single piezoelectric actuator,

    active flexures can be used in the transmission to create flapping asym-

    metries and, therefore, control torques. The hinges can be a lightweight

    replacement to additional actuators used for control, increasing the sys-

    tem lift-to-weight ratio while maintaining controllability [8].

    In this study, we propose to use shape memory polymer (SMP)-

    coated flexures as tunable stiffness joints. SMPs and shape memory

    alloys both fall into the category of smart materials and are capable of

    remembering and recovering an original shape after being deformed.

    In part with their shape-changing ability, SMPs also demonstrate large

    changes in elastic modulus when activated by external stimuli, such

    Manuscript received February 2, 2012; revised April20, 2012; accepted April24, 2012. Date of publication May 18, 2012; date of current version August 2,2012. This paper was recommended for publication by Associate Editor Y. Sunand Editor B. J. Nelson upon evaluation of the reviewers comments. The workof L. Hines was supported by the Department of Defense through the NationalDefense Science and Engineering Graduate Fellowship Program.

    L. Hines is with the Robotics Institute, Carnegie Mellon University,Pittsburgh, PA 15213 USA (e-mail: [email protected]).

    V. Arabagi and M. Sitti are with the Department of Mechanical Engi-neering, Carnegie Mellon University, Pittsburgh, PA 15213 USA (e-mail:[email protected]; [email protected]).

    Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

    Digital Object Identifier 10.1109/TRO.2012.2197313

    Fig. 1. (a) Single-wing flapping prototype with an SMP integrated activejoint. The flapping mechanism is mounted rigidly to a mechanical balance anda load cell for mean lift force measurement. Wing length is 20 mm. (b) Front-view photograph of the SMP and polyimide flexure. (c) Side-view conceptualdepiction of flexure layers.

    as direct heating or indirect heating including light, electric fields,

    and magnetic fields [9]. Because of their compact size and ease of

    integration, shape memory materials have found use as actuators in

    many miniature robotic platforms, including climbing [10], crawling

    [11], and rolling robots [12], as well as in miniature mechanisms, such

    as compact grippers [13] and devices geared toward minimally invasive

    surgical procedures [14][16]. Utilizing the stiffness change of a smartmaterial directly is less common; previously, SMPs have been used

    to create thermally changing microstructures [17], [18] and have been

    incorporated into composite cantilever beams [19], [20] to allow beam

    stiffness change.

    Here, we use a composite SMP structure to allow control of hinge

    stiffness and induce a change in functional behavior in a miniature

    robot while maintaining low system weight. Our testing platform was a

    single-wing system using a spherical four-bar transmission mechanism

    to amplify the motion of a bending bimorph piezoelectric actuator.

    The SMP-coated flexure was incorporated into one of the joints of the

    transmission to control the wing stroke amplitude and, thus, generated

    lift.

    II. EXPERIMENTS

    To fabricate the SMP integrated tunable stiffness flexure, as de-

    picted in Fig. 1(c), a layered carbon fiber and polyimide film (Kapton,

    Dupont) flexure was first manufactured according to the Smart Com-

    posite Microstructures methodology [4]. The Kapton serves as the

    flexible bending sheet connecting the two surrounding rigid carbon

    fiber layers of 60-m thickness (Torayca M60J). The Kapton bending

    sheet had dimensions of 3 mm (width) 0.75 mm (length) 6.5m

    (thickness) andwas bracketed by 2-mm-long carbon fiber sections. The

    epoxy SMP was mixed according to the process described in [21], with

    equal mass ratios of EPON 826 (Hexion), Jeffamine D230 (Huntsman),

    and Neopentyl glycol diglycidyl ether (TCI America) used. The for-

    mulation has a glass transition temperature of 50

    C, which is above

    1552-3098/$31.00 2012 IEEE

  • 8/14/2019 Shape Memory Polymer-Based Flexure Stiffness Control.pdf

    2/4

    988 IEEE TRANSACTIONS ON ROBOTICS, VOL. 28, NO. 4, AUGUST 2012

    ambient air temperature while remaining easily producible by a heat

    source. The mixture was applied by hand to one side of the manufac-

    tured flexure and thermally cured at 100 C for 1.5 h and postcured at

    130 C for 1 h. The final flexure, which is shown in Fig. 1(b), had a

    68-m layer of SMP on the Kapton bending sheet and weighed 4 mg.

    By creating a composite polyimide and SMP flexure, the mechani-

    cal strength and life-time of the hinge are increased significantly over

    the use of an only SMP-based flexure, which is necessary given thedesired repeated oscillatory loading. The problem of low mechanical

    strength of SMP materials, in general, is currently an active research

    topic with improvements demonstrated through both the use of various

    laminates [20], [22] and internal fillers for reinforcement [19].

    The cool and warm state bending stiffness of the flexure was mea-

    sured by applying an impulse and recording the response using a laser

    micrometer (Keyence, LS-3100) at room temperature and at approxi-

    mately 70 C, above the SMP transition temperature. An extension of

    known weight was added to amplify motion and make the system un-

    derdamped. At room temperature, the flexure had a rotational stiffness

    of 572 mNmm; when fully warmed, the stiffness was 11 mN mm.

    The flexure was incorporated into a rigidly mounted single-wing

    flapping prototype, as described in [8], with a passive wing flexure of

    rotational stiffness 11 mNmm. This system features the leading edge

    of the wing drivenby a bimorph piezoelectric actuator, whose motion is

    amplified by a spherical four-bar transmission mechanism. The trailing

    edge of the wing is allowed to passively rotate. The design of the piezo-

    electric actuator is critical to the lift generation in the flapping flight

    system to achieve high flapping amplitudes at sufficient wing flapping

    frequencies, more details of which can be found in [4], [6], and [8].

    The actuator here has a length of 18.5 mm which includes a passive

    rigid extension of 7.5 mm. From Laminate Plate Theory and following

    the work of Wood et al. [23], actuator bending stiffness is a predicted

    719 N/m with a blocking force of 84 mN at 200-V input. The original

    Kapton transmission flexure connecting the first transmission link to

    the body was replaced with the SMP integrated flexure, as pictured in

    Fig. 1(a). The system was mounted to a mechanical balance coupled toa load cell (30 g, Transducer Techniques) for measurement of average

    lift. Recorded lift data at 1000 Hz was sampled with a simple moving

    average filter with a length of 100 samples. Heat was applied to the

    flexure through an external heating element constructed from coiled

    nichrome wire (40 AWG, Newtons Third Rocketry). As the tempera-

    ture of the SMP itself cannot be taken directly without affecting system

    performance, the air temperature near the flexure was measured with

    a thermocouple. With an input of 2 W, air temperature 1 mm from the

    heating element reaches 50 C in less than 5 s and a maximum tem-

    perature of approximately 110 C after 15 s. Air temperature remains

    below 30 C at the piezoelectric actuator.

    As minimizing power consumption of the mechanism is of par-

    ticular importance to miniature robots and devices [24], the flexure

    dimensions were chosen to achieve nominal rotational joint motionwith no external heating applied and large deflections when warmed.

    Once incorporated into the transmission of the flapping mechanism,

    this allows the system to operate nominally at maximal lift with no

    additional power expenditure.

    Wing kinematics and mean lift of the test system were captured

    concurrently with a high-speed camera (pco.dimax) at 900 frames/s

    and load cell, as described previously, while the SMP flexure was

    heated beyond its transition temperature and then allowed to cool. The

    camera captured the system from its top view and was able to capture

    both the wing flapping angle (axis R1 in Fig. 1) and the wing rotation

    angle (axisR2 in Fig. 1). Wing kinematics were measured with point

    tracking in postprocessing. Power of 2 W was applied to the resistance

    heater, while the piezoelectric actuator was driven with a sinusoid

    0

    20

    40

    60

    80

    100

    120

    Angle(deg)

    Flapping Angle

    Rotation Angle (+)

    Rotation Angle ()

    0 5 10 15 20 25 30 350.5

    0

    0.5

    1

    Lift(mN)

    Time (sec)

    (a)

    (b)

    Fig. 2. (a) Wing peak-to-peak flapping angles and maximum rotation angles,and (b) lift of the prototype single-wing flapping mechanism over time withwarmed and cooled SMP integrated flexure. The system was driven at 36 Hzand 200-V peak-to-peak with 2 W inputted to the heating element at timesindicated with the red shaded area. Raw and filtered lift data are colored grayand black, respectively. Rotation angles are indicated with (+) and () forpositive and negative wing rotation about the nominal position since rotationwas not symmetric.

    of 36 Hz and 200-V peak-to-peak amplitude. Fig. 2 shows the system

    wing flapping and rotation angleswith changing lift over time. With thedescribed heating element, the system transitioned from a maximum

    lift of 0.52 mN to minimum lift of 0.1 mN in 3.5 s, a loss of 80%.

    Wing flapping angle decreased significantly, being 80 peak-to-peak

    when cool and dropping to 50 peak-to-peak when warmed (see video

    online [25]).

    The decrease in lift is caused by a combination of decrease in sys-

    tem resonant frequency and transmission displacement loss because

    of the heated, highly deformable flexure. With the cool SMP flexure,

    the maximum lift of 0.52 mN peaks at 36 Hz. When the flexure is

    heated, maximum lift is 0.25 mN and occurs at 28 Hz. Holding the

    driving frequency of the system constant accentuates the lift change. In

    Fig. 3(b), images 1 and 2 depict flexure behavior while both cool and

    warm. In both 1 and 2, a side view of the system and SMP flexure is

    illustrated, in which two images of maximum flexure deformation are

    superimposed. While cool, the flexure operates as a typical, albeit stiff,

    rotational flexure, leaving the transmission to function as originally

    designed. When heated, the flexure deforms significantly and allows

    translational motion between the fixed base and first transmission link,

    resulting in a loss of flapping amplitude.

    To produce maximal lift, ideally, the transmission links would func-

    tion as perfect rotational joints with no stiffness, resulting in ampli-

    fication of actuator motion without loss. Although the described cool

    SMP flexure is significantly stiffer than normal Kapton only flexures,

    the chosen transmission joint experiences only minimal rotational mo-

    tion, minimizing the effect of additional flexure stiffness. Indeed, with

    a replaced passive Kapton only flexure at the first transmission link,

    maximum lift was 0.46 mN and occurred at 36 Hz. Thesmall difference

  • 8/14/2019 Shape Memory Polymer-Based Flexure Stiffness Control.pdf

    3/4

    IEEE TRANSACTIONS ON ROBOTICS, VOL. 28, NO. 4, AUGUST 2012 989

    Side View2

    20 40 60 800

    0.1

    0.2

    0.3

    0.4

    0.5

    0.6

    Air Temperature (C)

    L

    ift(mN)

    1

    2

    Heating

    Cooling

    (a)

    SMP Transmission

    Flexure

    Heating Element

    1st Transmission

    Link

    Side View1(b)

    Fig. 3. (a) System lift change versus temperature for SMP integrated flexureheating and cooling. Each data point represents at least 15 s of captured data.(b) Photographs 1 and 2 depict side views of the SMP flexure when fully cooledand fully warmed, respectively. Two images are superimposed at the time ofmaximum flexure deformation with white dotted lines marking the position ofthe first transmission link.

    in maximum lift can be attributed to transmission misalignment which

    can occur when completely removing and replacing joints. Overall sys-

    tem lift is lower than what was seen in [8] due to the repaired passive

    wing flexure which is stiffer than optimal and asymmetrically rotates,

    as seen in Fig. 2; resonance and flapping amplitude between systems

    is comparable.

    To ensure that the flexure stiffness tuning is precisely controllable,

    system lift was recorded with discrete steps in the inputed power into

    the resistance heater. Fig. 3(a) depicts change in lift with change in

    air temperature. Each input power was held constant at least 15 s to

    ensure that both the temperature and lift stabilized. While transitioning,

    change in lift is almost linear with temperature, with little hysteresis.Such a resultdemonstrates that any lift valuebetweenthe fully cooledor

    heated state could be maintained with closed-loop temperature control.

    III. DISCUSSION

    While the current SMP flexure is heated with an external resistance

    heater, internal heating is possible, although with increased manufac-

    turing complexity. The current heater uses a little under 2 W of power

    to induce a full transition of the flexure; embedding nichrome wire

    into the SMP flexure layer would result in a more compact system

    and significantly reduce power expenditure necessary for a stiffness

    change. A single embedded wire along the edge of the flexible flexure

    sheet should not increase flexure stiffness significantly and allow faster

    SMP heating. Transition speed is the predominate concern for prac-

    tical use in control of a miniature flapping-wing-based flying robot.

    Currently, the fastest demonstrated change in lift is 0.12 mN/s, which

    may or may not be sufficient considering current flapping-wing con-

    trollers rely upon a per-wing stroke update frequency. The time lag

    for the flexure stiffness change, which is caused by the time to warm

    from room temperature, can be minimized with closed-loop tempera-

    ture control. Time to cool can be improved by using an SMP with a

    higher glass transition temperature, creating a larger gradient betweenSMP and room temperature.

    Although the presented controllable flexure design relies on non pin

    joint behavior of a rotational flexure in its warm state, it is possible to use

    the same basic strategy to create a controllable purely rotational joint.

    With modified flexure dimensions, higher aspect ratios, and thinner

    SMP coatings, the flexure can be constrained to rotational motion,

    while still allowing changes in stiffness. Designers, however, should

    be wary of fatigue failure under cyclic loading. Frequent flexure failure

    was observed when the active flexure was incorporated into joints with

    higher angular displacements (up to 100), such as the wing rotational

    flexure or the final transmission link, which directly drives the wing

    stroke angle. With large deformations, failure would occur after 5 s

    of operation or 200 cycles. Placement on a transmission joint with

    low required angular displacements eliminated this problem.

    IV. CONCLUSION

    In this study, we have demonstrated an SMP-integrated flexure with

    tunable stiffness incorporated into a single-wing flapping-based flying

    robot prototype. The flexure allows lift change without modification

    of the piezoelectric actuator driving waveform, opening the possibility

    for wing force asymmetry without additional piezoelectric actuators

    in multiwing systems. As a future work, heating elements will be

    embedded into or close to the flexures to decrease power consumption

    and to decrease thestiffness change response time forhigherbandwidth

    motion control. Such active stiffness tunable flexure joints could be

    applied to any flexural miniature mobile robot and device mechanisms.

    ACKNOWLEDGMENT

    The authors would like to thank the members of the NanoRobotics

    Laboratory for all their support and suggestions.

    REFERENCES

    [1] N. Lobontiu,Compliant Mechanisms: Design of Flexure Hinges. BocaRaton, FL: CRC, 2002.

    [2] S. Zelenika, M. G. Munteanu, and F. D. Bona, Optimized exural hingeshapes for microsystems and high-precision applications, Mech. Mach.Theory, vol. 44, pp. 18261839, 2009.

    [3] J. P. Whitney and R. J. Wood, Aeromechanics of passive rotation inflapping flight, J. Fluid Mech., vol. 660, pp. 197220, 2010.

    [4] R. J. Wood, The first takeoff of a biologically inspired at-scale roboticinsect, IEEE Trans. Robot., vol. 24, no. 2, pp. 341347, Apr. 2008.

    [5] B. M. Finio and R. J. Wood, Distributed power and control actuation inthe thoracic mechanics of a robotic insect, J. Bioinspirat. Biomimet.,vol. 5, p. 045006, 2010.

    [6] M. Sitti, Piezoelectrically actuated four-bar mechanism with two flexi-ble links for micromechanical flying insect thorax, IEEE/ASME Trans.

    Mechatronics, vol. 8, no. 1, pp. 2636, Mar. 2003.[7] X. Deng, L. Schenato, and S. Sastry, Flapping flight for biomimetic

    robotic insects Part II: Flight control design, IEEE Trans. Robot., vol.22,no. 4, pp. 789803, Aug. 2006.

    [8] V. Arabagi, L. Hines, and M. Sitti, Design and manufacturing of a con-trollable miniature flapping wing robotic platform, Int. J. Robot. Res.,

    vol. 31, no. 6, pp. 785800, May 1, 2012.

  • 8/14/2019 Shape Memory Polymer-Based Flexure Stiffness Control.pdf

    4/4

    990 IEEE TRANSACTIONS ON ROBOTICS, VOL. 28, NO. 4, AUGUST 2012

    [9] P. T. Mather, X. Luo, and I. A. Rousseau, Shape memory polymer re-search, Annu. Rev. Mater. Res., vol. 39, pp. 445471, 2009.

    [10] J. Koh, S. An, and K. Cho, Finger-sized climbing robot using artificialproleg, inProc. Int. Conf. Biomed. Robot. Biomechatron., Tokyo, Japan,2010, pp. 610615.

    [11] B. Kim, M. G. Lee, Y. P. Lee, Y. Kim, and G. Lee, An earthworm-likemicro robot using shape memory alloy actuator, Sens. Actuat. A: Phys.,vol. 125, no. 2, pp. 429437, 2006.

    [12] Q. Chang-jun, M. Pei-sun, and Y. Qin, A prototype micro-wheeled-robotusing SMA actuator, Sens. Actuat. A: Phys., vol. 113, no. 1, pp. 9499,2004.

    [13] Z. W. Zhong and C. K. Yeong, Development of a gripper using SMAwire, Sens. Actuat. A: Phys., vol. 126, no. 2, pp. 375381, 2006.

    [14] V. R. C. Kode and M. C. Cavusoglu, Design and characterization of anovel hybrid actuator using shape memory alloy and DC micromotor forminimally invasive surgery applications, I EEE/ASME Trans. Mecha-tronics, vol. 12, no. 4, pp. 455464, Aug. 2007.

    [15] H. M. Wache, D. J. Tartakoska, A. Hentrich, and M. H. Wagner, De-velopment of a polymer stent with shape memory effect as a drugdelivery system, J. Mater. Sci.: Mater. Med., vol. 14, pp. 109112,2003.

    [16] W. G. Bae, J. H. Choi, S. H. Lee, D. Kang, K. W. Jung, and K. Y. Suh,Centering mechanism for micro vessel robot using micropatterned shapememory polymers, in Proc. Int. Conf. Biomed. Robot. Biomechatron. ,Tokyo, Japan, 2010, pp. 594598.

    [17] S. Kim, M. Sitti, T. Xie, and X. Xiao, Reversible dry adhesives withthermally controllable adhesion, Soft Matter, vol. 5, pp. 36893693,2009.

    [18] E. Manias, J. Chen, N. Fang, and X. Zhang, Polymeric micromechanicalcomponents with tunable stiffness, Appl. Phys. Lett., vol. 79, pp. 17001702, 2001.

    [19] Y. Liu, K. Gall, M. L. Dunn, and P. McCluskey, Thermomechanics ofshape memory polymer nanocomposites, Mech. Mater., vol. 36, no. 10,pp. 929940, 2004.

    [20] C. Zhang and Q. Ni, Bending behavior of shape memory polymerbased laminates, Composite Struct., vol. 78, no. 2, pp. 153161,2007.

    [21] T. Xie and I. A. Rousseau, Facile tailoring of thermal transition tem-peratures of epoxy shape memory polymers, Polymer, vol. 50, no. 8,pp. 18521856, 2009.

    [22] Z. Li and Z. Wang, Characterization of the flexural behavior ofSMP sandwich beam, Adv. Mater. Res., vol. 123125, pp. 939942,2010.

    [23] R. J. Wood, E. Steltz, and R. S. Fearing, Optimal energy density piezo-electric bending actuators, Sens. Actuat. A: Phys., vol. 119, no. 2,pp. 476488, 2005.

    [24] M. Sitti, Miniature devices:Voyage of themicrorobots,Nature, vol. 458,pp. 11211122, 2009.

    [25] (2012). [Online]. Available: http://nanolab.me.cmu.edu/projects/FlappingRobot/