1

IAC-04-IAF-I.3.B

WaalBots for Space Applications

Carlo Menon1, Michael Murphy2, Francesco Angrilli1, and Metin Sitti2 1CISAS "G. Colombo”, University of Padova, Italy

2Carnegie Mellon University, Mechanical Engineering Department, PA 15213, USA (menoncarlo@stargatenet.it, mikemurphy@cmu.edu, francesco.angrilli@unipd.it, sitti@cmu.edu)

Recent progress in the understanding of lizard locomotion has motivated the design of small robots with biomimetic properties that provide an inherent advantage for space applications. Like lizards, these revolutionary small space robots (WaalBots), will exploit Van der Waals forces for mobility and adhesion on spacecrafts and planet surfaces. Dry adhesion allows lizards to rapidly and efficiently scale walls and to hang from ceilings with an attachment mechanism that inherently requires low power consumption. In this paper, a comprehensive description of unique characteristics of WaalBots is presented. Latest achievements and methods of fabricating bio-mimetic dry adhesives along with force measurement results are described. Design, analysis, construction and tests of climbing WaalBot prototypes are also presented.

1 INTRODUCTION A robot which can operate on a vertically

oriented surface allows the possibility of automating tasks which are currently accomplished manually, affording an extra measure of human safety, often in a more cost effective manner. Some wall-climbing robots are in use in industry today cleaning high-rise buildings and performing inspections in dangerous environments such as storage tanks for petroleum industries and nuclear power plants 1. Recently, there has also been interest in using robots to inspect and repair space vehicles.

In literature, two main types of attachment mechanisms were studied and developed for wall climbing robots. The most common type is suction adhesion 2-4 where the robot carries an onboard pump to create a vacuum inside cups which are pressed against the wall or ceiling. This type of attachment has some major drawbacks associated with it. The suction adhesion mechanism requires time to develop enough vacuum to generate sufficient adhesion force. This delay reduces the speed at which the robot can climb. Another issue associated with

suction adhesion is that any gap in the seal can cause the robot to fall. This drawback limits the suction cup adhesion mechanism to relatively smooth, non-porous, non-cracked surfaces. Lastly, the suction adhesion mechanism relies on the ambient pressure to stick to a wall, and therefore is not useful in space applications as the ambient pressure in space is essentially zero.

Another common type of adhesion mechanism is magnetic adhesion 5,6. Magnetic adhesion has been implemented in wall climbing robots for specific applications such as nuclear facilities inspection. In specific cases where the surface allows, magnetic attachment can be highly desirable for its inherent reliability. Despite its advantages, magnetic attachment is useful only in specific environments where the surface is ferromagnetic, so for most applications it is an unsuitable choice.

In this paper an unconventional attaching method is discussed. Inspired by climbing animals, new strategies are developed for use in climbing robots. Many animals have the desirable ability to stick to and climb on various surfaces. Insects, beetles, skinks, anoles, frogs and geckos have been studied for their sticking abilities. In particular, geckos are the most

2

interesting because of their size. The Tokay Gecko can weigh up to 1 kg and reach lengths of 35 cm yet is still able to run inverted and cling to smooth walls. By studying and imitating the attachment mechanism of this gecko, a new generation of space robots can be developed making locomotion possible on almost any kind of surface without contaminating the surrounding environment.

2 DRY ADHESION For over 2 millennia, humans have watched

lizards and bugs scale vertical surfaces in awe 7. Only recently has the attachment mechanism of these animals been understood. It is now possible to use similar mechanisms to allow robots to climb in the same manner as these animals.

Geckos’ ability to climb surfaces, whether wet or dry, smooth or rough, has attracted scientists attention for decades. By means of compliant micro/nano-scale high aspect ratio beta-keratin structures at their feet, geckos manage to adhere to almost any surface with a controlled contact area 7. It has been shown that adhesion is mainly due to molecular forces such as van der Waals forces 8.

The gecko’s ability to stick to surfaces lies in its feet, specifically the very fine hairs on its toes. There are billions of these tiny fibers which make contact with the surface and create a significant collective surface area of contact. The hairs have physical properties which let them bend and conform to a wide variety of surface roughness, meaning that the adhesion arises from the structure of these hairs themselves.

The structure of the biological gecko foot-hair is very complicated and miniscule. Each hair-like fiber is made from multiple sections. Each fiber consist of a micro-hair (seta) which is roughly 5 microns in diameter, and atop each of these micro-fibers sit hundreds of nano-fibers (spatulae) which are 200 nanometers in diameter. There are between 100 and 1,000 spatulae on the end of each seta 9.

Although the surface area of each of the hair

tips is very small, the combination of the area of billions of these hairs makes the effective surface area quite large, and surface forces (particularly van der Waals forces) become significant. Since the hairs are individually compliant, they can deform to match different surface roughness. Also, because of their hydrophobic nature, the gecko fibers are self-cleaning. In order for the dry adhesion to function, a small preload force normal to the surface is required to force the compliant hairs to configure themselves properly. Once this preload force has been applied, the material will stick until peeled off. The adhesion force can be as high as 10N per 1cm2 area 7.

Since dry adhesion is caused by van der Waals forces, surface chemistry is not of great importance 8. This means that dry adhesion will work on almost any surface.

Much like the real gecko material, the synthetic adhesive will be super-hydrophobic and therefore be self-cleaning allowing for long lifetime robots. The nature of the adhesion force is such that no energy is required to maintain attachment after it has been initiated, so a robot using dry adhesion could hang on a wall indefinitely with no power consumption.

Another benefit of dry adhesion is the speed at which attachment and detachment is possible. The attachment is nearly instantaneous as is the detachment. They both only depend on the force applied. This allows for no delay in locomotion, thus very fast locomotion speeds. Furthermore, it is not necessary to time the attachment as critically as with the electromagnetic attachment, only a force is required, so the attachment is passive in nature, and therefore simple to control.

Dry adhesion is more robust than the suction adhesion mechanism, if the dry adhesion pad encounters a crack or gap, there will still be adhesion on the parts of the pad that have made contact. This behavior permits a robot using dry adhesion to climb on a wider variety of surfaces. Also, since dry adhesion does not rely heavily on the surface material or the atmosphere, it is suitable for use in the vacuum of space as well

3

as inside liquid environments

3 SYNTHETIC HAIR FABRICATION As a first step in the synthetic gecko fiber

fabrication it is necessary to develop techniques to create the micro and nano-fibers independently. Once this is accomplished, it is possible to begin integrating the two types of fibers into a single process. The final structure will be a micro-fiber with nano-fibers branching out the end of the micro-fiber as seen in Fig. 1.

In this paper, two methods to replicate the structure which use micro-molding techniques are described whereas the theoretical aspects are discussed in 9,10. The first fabrication method utilizes commercially available components while the second method utilizes MEMS techniques to fabricate custom master molds. In both methods, liquid polymer is poured over the molds and cured. The cured molded polymer emerges in the desired physical form. It is possible to approximate the physical characteristics of the beta keratin by selecting the proper polymer. In the first method a commercially purchased porous membrane is used as the mold. This membrane can have pore size of 0.02-20µm, thickness of 5µm, and pore density of 105-108 pores/cm2. The larger sized diameter polycarbonate membranes have a random orientation of the nano-pores (�15°) created by a nuclear track etch while the smaller alumina membranes have very high density and directional uniformity. The thin porous membrane is attached to a substrate and is molded with polydimethyl siloxane (PDMS) or a similar polymer under vacuum. The vacuum is used to evacuate air from the pores so that the polymer can easily flow into them. After curing, the hardened polymer can either be mechanically peeled away from the membrane or the membrane can be chemically etched away. Results from this method are promising. 200nm diameter high aspect ratio fibers have been produced (Fig. 2), which are similar to the distal hairs found in geckos. From this image it is clear that there is bunching or matting occurring between the fibers, likely due to the

superfluous length of the fibers. As the length of the fibers increases, the inter-fiber adhesion force surpasses the spring force of the fiber to remain upright and the fibers begin to bunch.

Fig. 1Micro and nano-scale gecko foot-fibers

(Schematic)

This problem is caused by the high aspect

ratio of the commercially available nanopore membrane pores as well as their high density. To avoid this matting issue, a second method of fabrication was developed in which the density, diameter and length could be independently controlled. This method entails patterning a silicon wafer through photolithography and using a deep reactive ion etch to create a negative mold for the fibers. As a last step in the etching process a thin conformal layer of fluorocarbon is deposited. This layer reduces the adhesion between the polymer and the mold which decreases the chances of the fibers breaking off and remaining in the mold during the peeling process, increasing yields. The same vacuum molding and heat curing process is used to create the fibers and the molded polymer is mechanically peeled off of the silicon. Micro-fiber results from this process can be seen in Fig. 3.

4

Fig. 2 200nm diameter high aspect ratio polymer nano-fibers

Fig. 3 4µm diameter polymer micro-fibers

These 4µm diameter fibers show a very

uniform structure, demonstrating the advantage of this new technique. Previous problems with fibers breaking off during the peeling phase seem to have been alleviated with the use of the fluorocarbon passivation layer, as it is rare to find any broken fibers.

4 WAALBOTS DESIGN AND CONSTRUCTION

In this section four robot prototypes are discussed. In particular sub-section 4.1 deals with two wheeled robots, whereas section 4.2 presents two robots inspired by the Gecko’s gait.

4.1 Wheeled Robots

A climbing robot design using dry adhesion

forces has to be developed in order to maximize the effectiveness of the attachment system. In particular there are three main requirements for developing such a robot:

1. Maximize the attachment area. 2. Apply preload between robot and

vertical surface to increase the attachment force.

3. Use a peeling motion during the detaching phase.

Two different vehicle concepts were developed. The first one is a wheg (wheel-leg) vehicle that uses legs with adhesive feet for climbing vertical surface. The second one is a tread-based locomotive mechanism using a rubber belt in place of a chain tire.

4.1.1 Robot Design

To optimize the design of Waalbot, detailed analysis was employed. Of particular importance is the analysis of the robot’s “tail” (Fig. 4). Its purpose is to preload the adhesive against the climbing surface.

Fig. 4 Forces and dimensions of the climbing

robot model used on FEM optimization

Finite Element Methods (FEM) was used to solve the over-constrained model in order to optimize the tail properties and the position of the center of mass. In the FEM simulation, the climbing robots were modeled as three beam

5

elements having null masses. The gravitational force was applied in the center of mass of the system.

Fig. 4 shows the directions of the forces that act on the system. The optimization focused on minimizing the force R10 (see Fig. 4) since that corresponds to the attaching requirements for the employed adhesive. The results of the optimization, presented in Fig. 5 and Fig. 6, correspond to a vehicle having the same dimensions of the tank robot. Specifications of the two prototypes are shown in Table 1.

Property Units Tank Whegs Voltage V 2-6 2-6

Weight kg 0.1 0.09 Length* M 0.1 0.06 Width m 0.07 0.08 Height m 0.05 0.06

Tail length m 0.09 0.09 *Without tail

Table 1 Specifications of Tank and Wheg robots

Fig. 5 shows how the force R10 varies when changing the length and rigidity of the tail. The attaching force has a monotone behavior with respect to the Young’s modulus but there is a local minimum for the tail length. From these results, the optimal tail length for the current configuration should be 0.12 meter long and the Young’s modulus should be the highest possible.

Fig. 5 Adhesive force R10 required by a climbing robot as a function of tail length and stiffness

The position of the center of mass is also of great interest for a climbing vehicle. Taking into account the sketch depicted in Fig. 4, the performed simulations were conducted changing the angle γ and keeping unchanged the variables H, L1 and L2.

Fig. 6 Adhesive force required by a climbing robot as a function of center of mass position

The results of Fig. 6 show that the center of mass should be positioned on the fore part of the vehicle for reducing the attaching force R10. Nevertheless, since from 20° to 70° the function does not present a steep slope, small changes in positions of the center of mass do not affect the performance of the vehicle.

4.1.2 Construction and Tests

Two wheeled vehicles were developed to show the feasibility of climbing mechanisms. The first one, shown in Fig. 7, uses legged wheels. This robot has 3 feet per wheg. Each foot has one degree of freedom through a passive revolute joint so that it is able to maintain contact with the surface throughout 120° of leg rotation. Conventional adhesive pads were placed on the bottom surface of these feet. To ensure that the feet were properly aligned as they approach the surface, elastic material was used as a spring to pull the feet back into the proper position after each detachment.

For proof of concept experiments, a double-sided foam adhesive was used to demonstrate the locomotion feasibility. The robot was able to

6

climb a smooth acrylic surface in vertical and past vertical climbing angles.

Fig. 7 Robot with legged wheels: CAD design (left image) and photo of the real prototype (right image)

The robot detached from the wall on occasion, however the problems seemed to be associated with the adhesive pads and robot construction rather than the robot design. In the current design both whegs (wheg=wheel+leg) rotate in phase on the same shaft. This allows for straight line motion only, which is very limiting. However, this design can be modified to allow for independent wheg rotation. When properly implemented, this will enable the robot to make turns and increase mobility.

Fig. 8 Photo of the tread-based locomotion

mechanism

The second robot, depicted in Fig. 8, was a tread vehicle with customized tires. The chassis, pre-tensioning system and wheels were fabricated by means of a three dimensional prototyping machine. Different types of tires were built for testing the performance of the mechanism. In particular two were of great interest. The first type used foam of high-stickiness that allowed the robot to climb vertical surface. The vehicle was also able to climb surfaces inclined at 110° with respect to the horizontal ground plane. Since the robot is intended to utilize the synthetic dry adhesive as attachment mechanism, customized tires were designed and molded using PDMS, the same material that was chosen for the synthetic hair fabrication. For the sake of simplicity, the treads tires were built without hairs since the feasibility of the vehicle was the first goal to achieve.

The wireless vehicle, equipped with PDMS treads, was able to climb an acrylic surface sloped at 75° showing a reliable behavior. In Fig. 9, the temporal evolution of the power consumption of the robot is presented for several sloped surface and voltage inputs. At maximum speed, corresponding to an input of 6V, the vehicle consumes an average power of 1.1 W.

Comparing the three graphs of Fig. 9, the power consumption of the climbing robot increases when the slope of the inclined surface is increased. The power measurements also show microscopic stick-slip behavior that was not noticed with visual macroscopic inspection. In fact, for high voltage input, Fig. 9c presents lower peaks respects to the other two figures. This behavior is associated to stick-slip phenomena that are more significant when the elastic belts of the tread-robot are stretched. One possible solution for this issue is to increase the stiffness of the tires.

Equipping the tread-vehicle with synthetic hair would greatly increase the performance of the robot making it capable of climbing many types of surfaces for any slope angle without contaminating the surrounding environment.

7

(a)

(b)

(c)

Fig. 9 Temporal evolution of power consumption of the tread-robot moving on: (a) 0°, (b) 40°, (c) 75° sloped surfaces

4.2 Gecko Robots

The Gecko lizard, known for its climbing abilities, has inspired the kinematics and locomotion system of two climbing robot prototypes shown in

Fig. 10. Unique features of these novel Gecko

inspired robots include: • High maneuverability, speed, and agility

due to fast attachment and detachment in

any orientation; • Possibility of carrying higher payloads (a

Gecko can carry a payload the same as its body weight during vertical wall-climbing);

• Accessing small and difficult to reach areas;

• Generating high attachment forces for realizing mechanical work during space robotic mission especially for maintenance applications;

• Autonomous and on-line monitoring, inspection and maintenance of surfaces by integrated sensing and manipulation tools.

The two developed Gecko robots differ by the technologies employed. The first robot, called Rigid Gecko Robot (RGR), uses conventional motors for reliability and power, enabling the possibility of future repairing operations in space servicing missions. The second robot, called Compliant Gecko Robot (CGR), is a research platform to develop technologies for future miniature climbing robot. This design uses Shape Memory Alloy wires (SMA) and a flexible spine for mimicking gecko gait.

Fig. 10 Pictures of Gecko Robot prototypes: Rigid Gecko Robot (RGR) (left photo) and Compliant Gecko Robot (CGR) (right photo).

The Gecko kinematic data obtained by 11 were

implemented and modeled to simulate the two-dimensional motion of the gecko animal. These simulations suggested the following characteristics for a climbing robot:

1. Center of mass close to the climbing

Foot Compliant back

Foot motor

8

surface, 2. Very light structure and light locomotion

systems, 3. Reliable robotic system especially during

climbing phases. The strategy was to develop a robot with only

necessary degrees of freedom to climb. The robot was thus not made heavier by auxiliary motors and sensors that could compromise its climbing performances. Particular emphasis was given to the design of an effectiveness kinematics and to the optimization of the robot’s geometry.

4.2.1 Robot Construction

The fabrication approach for the RGR and CGR were different. For the first robot, conventional materials and fabrication techniques were used since RGR was conceived for operating in a space environment. The material chosen for the chassis of the RGR was aluminum alloy. Starting from sheets of metal, the robotic frame was fabricated by means of a folding technique. Miniature DC motors were used as actuators. The motor torque was amplified by a 81:1 gearbox and the motor output 25mNm torque when operated at 5V. Each motor was equipped with a torque spring able to return the motor’s shaft to its initial position when the input was 0V. Mechanical end-stops limited the shaft rotation range to 60°. By means of torque springs and end-stops, the necessity of position sensors is avoided.

Fig. 11 Control Strategy for one robot step. The rotational position of each joint versus the time is represented.

Dampers mounted next to the end-stops were employed for limiting vibrations caused by impact during leg movement. The RGR is controlled by a PIC 16F877 micro-controller integrated in a custom circuit board. The PIC controlled and synchronized the movements of both the legs and the back of the RGR. Fig. 11 shows the control strategy used for one-full step.

The CGR uses Flexinol SMA wires with a diameter of 50µm and a transition temperature of 90°C (High Temperature SMA wires). Several thin wires were used instead of few thick wires to increase the cooling rate. The wire length for the compliant body was about 10 cm and the external power system supplied 5V for actuation. The thermal cycle rate was about 1cyc/s. The compliant body of the CGR was built with a composite structure composed of three layers with the following layout:

• Unidirectional prepreg glass fiber (S2Glass) with a thickness of 30µm

• Prepreg carbon fiber (M60J) weaves with a thickness of 80µm

• Unidirectional glass fiber (S2Glass) with a thickness of 30mm

Properties of these materials are shown in Table 2.

Material M60J S2Glass E1 (GPa) 350 60 E2 (GPa) 350 7 G12 (GPa) 7 7

Table 2 Properties of the layers used for the robot’s composite back

Composite laminate theory 12 was used and the properties of the laminate resulted:

E1 (GPa) E2 (GPa) G12 (GPa) ν12 226 205 7 0.3

Table 3 Properties of the robot composite laminate

Using six 50µm diameter SMA wires for each side of the compliant back and a laminate width of 24mm, the compliant back was structurally verified. Three failure theories were used: Tsai-Hill, Hoffman, and Tsai-Wu 12. Since all of them

9

showed a failure factor greater than one, the compliant back was suitable for provide the locomotion of the CGR.

The use of Glass fiber had two different purposes: to reinforce the compliant body structure; to electrically isolate the frame from the SMA wires when these were in contact during the robot’s back deflection. For improving the isolation, a layer of epoxy was spread out in order to coat the compliant back. A spinner machine was used to obtain a thin layer of coating for keeping light the composite chassis of the robot. Jump cables were used to supply the power requested by the SMA wires. For this purpose, silver wires coated with Teflon and having a diameter of 140µm were used in order to limit mechanical disturbances to the CGR during its motion.

4.2.2 Tests and Results

As regards the RGR, it had a robust behavior while walking in an horizontal plane showing a gait movement similar to the real gecko animal. This robot was able to climb an acrylic surface with a slope of 65° respect the horizon. The performance of the robot, which potentially was able to climb a vertical surface, was mainly limited by the absence of encoders for the feedback of the robotic legs’ positions. The detaching phase was thus not controlled increasing the possibility of failure during the climbing phase. The RGR had a weight of about 80g, a length of about 0.1 m and a width of about 0.1 m. In Fig. 12 a snap shot sequence of the RGR is shown during its climbing phase.

Fig. 12 Snap shots of the RGR during climbing phase

The behavior of CGR’s compliant back was

characterized by means of static and dynamic tests. The measurement setup included a laser scan micrometer able to measure the displacement of the compliant back during SMA wires’ contraction. The resolution of the micrometer was 2µm. The length of the compliant back was of 9cm. From static measurements the function between the voltage applied to the SMA wires and the displacement of the compliant back was obtained as shown in Fig. 13. A model of the robot based on these experimental results can be implemented in a feed-forward loop to precisely control the movements of the robot.

The dynamic behavior of the compliant back was characterized by testing it under dynamic loads. Its displacements were recorded every 2.5ms during the SMA wires’ contraction. Fig. 14 shows the temporal evolution of the compliant back for both heating and cooling SMA phases using three different voltages. Cycling the contraction and release of the SMA wires continuously, a maximum frequency of 1hz is realized. Increasing the voltage applied to the SMA wires results in three differences: increasing the voltage from 3V to 6V the maximum displacement barely increased; by increasing the voltage, the heating phase was more rapid but since cooling is the limiting factor in increasing the cycle speed, this speed increase has negligible effects; increasing the voltage caused a jitter effect. These dynamic characteristics suggested the use of the lowest voltage required for a desired displacement.

10

Fig. 13 Behavior of the CGR back during SMA wires’ contraction

In addition, using 4V instead of 6V, the power consumption of the CGR is reduced. When using 5V (Fig. 14), an interesting behavior occurs: the first pick is lower than the second one. This shows instability between the SMA wires and the elastic compliant back. While the SMA wires were contracting, the compliant back was accelerated. Its inertia caused a temporary overcoming of the expected displacement and, at the same time, a relaxation of the wires. This effect can be reduced decreasing the mass and increasing the damping of the compliant back. This can be achieved reducing the epoxy matrix of the laminate and using aramidic fibers instead of carbon fibers.

Fig. 14 Dynamic behavior of SMA wires using 4V (top), 5V (middle), and 6V (bottom)

For speed and reliability, the CGR used four

DC motors to lift its legs. Nevertheless this choice compromised the performance of the

CGR since the motors added significant mass (28g) to the lightweight robot (35g). The CGR was able to climb a 60° acrylic surface mimicking the gecko gait. Fig. 15 shows the CGR while climbing.

Fig. 15 Snap shots of the RGR during climbing phase

5 CONCLUSION The importance of realizing mechanisms able

to climb every kind of surface without contaminating the surrounding environment has driven the research to focus on the ability of animals to climb vertical walls of space infrastructure. New techniques for fabricating synthetic micro-fibers for use as dry adhesives and the results of this process were presented. Four robotic prototypes, equipped with conventional adhesives, are also shown and discussed. The prototypes were able to climb vertical smooth surfaces, demonstrating the feasibility of the novel robot designs. Future work includes improving the synthetic hair fabrication and the implementation of this material in robots for space operations.

ACKNOWLEDGMENT The authors thank all members of the

Carnegie Mellon University NanoRobotics Laboratory for their support.

REFERENCES 1. W. Yan, L. Shuliang, X. Dianguo, Z. Yanzheng,

S. Hao, and G. Xueshan, “Development and application of wall-climbing robots,” Proceedings of the IEEE International Conference on Robotics and Automation, pp 1207-12, 1999.

2. Pack, R.T.; Christopher, J. L.; Kawamura, K.; “A Rubbertuator-Based Structure-Climbing

Instable behavior

11

Inspection Robot,” Proceedings of the IEEE International Conference on Robotics and Automation, pp. 1869-1874, 1997.

3. T. Yano, T. Suwa, M. Murakami, and T. Yamamoto, “Development of a Semi Self-Contained Wall Climbing Robot with Scanning Type Suction Cups”, Proc. of the Intelligent Robot Systems Conference, pp. 900-905, 1997.

4. A. Nagakubo and S. Hirose, “Walking and running of the quadruped wall-climbing robot”, Proceedings of the IEEE International Conference on Robotics and Automation, pp. 1005-1012, 1994.

5. H.R. Choi, S.M. Ryew, T.H. Kang, J.H. Lee, and H.M. Kim, “A wall climbing robot with closed link mechanism”, Proc. of the Intelligent Robotic Systems Conference, pp. 2006-2011, vol. 3, 2000.

6. L. Shuliang, Z. Yanzheng, G. Xueshan, X. Dianguo, and W. Yan, “A wall climbing robot with magnetic crawlers for sand-blasting. Spray-Painting and Measurement,” High Technology Letters, pp 86-88, vol. 10, 2000.

7. K. Autumn, Y. Liang, T. Hsieh, W, Zesch, W.P. Chan, T. Kenny, R. Fearing, and R.J. Full, “Adhesive force of a single gecko foot hair,” Nature, 405, pp. 681-5, 08 June 2000.

8. K. Autumn, M. Sitti, Y.A. Liang, A.M. Peattie, W.R. Hansen, S. Sponberg, T. Kenny, R. Fearing, J.N. Israelachvili, and R.J. Full, “Evidence for van der Waals adhesion in gecko state,” Proceedings of the National Academy of Sciences, vol. 99, pp. 12252-6, 2002.

9. G. Shah and M. Sitti, "Modeling and Design of Biomimetic Adhesives Inspired by Gecko Foot-Hairs," Proc. of the IEEE Conf. on Robotics and Biomimetics, Shenyang, China, Aug. 2004 (to appear).

10. M. Sitti and R. S. Fearing, "Synthetic Gecko Foot-hair Micro/Nano-Structures as Dry Adhesives," Journal of Adhesion Science and Technology, vol. 17, no. 8, pp. 1055-1074, 2003.

11. A. Zaaf, R. Van Damme, A.Herrel and P.Aerts, “Limb joint Kinematics during vertical climbing and level running in a specialist climber: Gekko gecko Linneus, 1758 (Lacertilia: Gekkonidae)”, Belg. J. Zool., pp 173-183, vol.131, 2001.

12. Isaac M. Daniel, Ori Ishai, “Engineering Mechanics of Composite Materials”, Oxford Publishers, 1994.