You are on page 1of 6

2012 IEEE International Conference on Robotics and Automation RiverCentre, Saint Paul, Minnesota, USA May 14-18, 2012

Design Considerations for Attachment and Detachment in Robot Climbing with Hot Melt Adhesives
Liyu Wang, Fabian Neuschaefer, Remo Bernet and Fumiya Iida
Abstract Robust climbing in unstructured environments is a long-standing challenge in robotics research. Recently there has been an increasing interest in using adhesive materials for that purpose. For example, a climbing robot using hot melt adhesives (HMAs) has demonstrated advantages in high attachment strength, reasonable operation costs, and applicability to different surfaces. Despite the advantages, there still remain several problems related to the attachment and detachment operations, which prevent this approach from being used in a broader range of applications. Among others, one of the main problems lies in the fact that the adhesive characteristics of this material were not fully understood n the context of robotic climbing locomotion. As a result, the previous robot often could not achieve expected locomotion performances and contaminated the environment with HMAs left behind. In order to improve the locomotion performances, this paper focuses on attachment and detachment operations in robot climbing with HMAs. By systematically analyzing the adhesive property and bonding strength of HMAs to different materials, we propose a novel detachment mechanism that substantially improves climbing performances without HMA traces.

I. INTRODUCTION There has been an increasing interest in using adhesive materials for robotic climbing [1-5]. In contrast to prior technical solutions such as by using vacuum suction [6], magnetic attachment [7] or gripping mechanisms [8], adhesion-based approach does not require continuous energy supply (as with vacuum suction) or pre-dened structures (i.e. grippers or ferromagnetic surfaces) to maintain a robot on a surface. However, most adhesive materials have a relatively low attachment strength (103 -104 Pa), and some materials lose their adhesiveness after being used for a limited time. In our lab we are developing climbing robots based on hot melt adhesives (HMAs). HMAs have a number of interesting properties: they are adhesive uids with a low bonding strength at a high temperature, while solid with a high bonding strength at room temperature; they can be repeatedly transformed between uid and solid phases by controlling their temperature; and the material has been used in many industrial applications and proved to be economical. By exploiting these properties, we have previously prototyped a climbing robot that demonstrated successful steps on a vertical surface [9]. The advantage of the approach lies in its high attachment strength (105 -106 Pa), and that it can be applied to different surfaces. However, HMAs adhesive
This work was supported by Swiss National Science Foundation, Grant No. PP00P2123387/1. All authors are with Bio-Inspired Robotics Laboratory, Department of Mechanical and Process Engineering, ETH Zurich, 8092, Zurich, Switzerland [liyu.wang, fumiya.iida]@mavt.ethz.ch,

properties were not fully understood, and the attachment and detachment operations during climbing were done in a naive way. For example, the robot used a large amount of HMAs to achieve a strong attachment, while detachment was realized by heating up the HMA to a high temperature so that it became uidic and adhesive. There are several negative consequences from this. First, since the HMA became adhesive during detachment, a trace was left on the climbing surfaces. Second, due to the continuous loss, new HMAs had to be supplied for subsequent steps, which not only increased design and control complexity, but also required longer time to climb. Third, over-use of HMAs cost more effort for the robot to detach itself. All those problems are critical for a broader use of the HMA-based method, and considerations for attachment and detachment operations are signicant from this aspect. The goal of the paper is to consider the design for better attachment and detachment operations in the HMA-based climbing approach. By exploiting HMAs adhesive properties and bonding strength to various materials, we propose a novel detachment mechanism for the robot not to leave a trace on the climbing surfaces. Under this detachment mechanism, we conduct systematic investigations on handling of HMAs for climbing locomotion. With a set of feasible models that describe bonding force and bonding moment, we show how an allowable amount range of HMAs can be determined from the climbing requirement, and how attachment can be made stronger with minimal HMAs in the range. We further implement the design considerations in a new climbing robot, which demonstrates faster climbing on a vertical surface without leaving an HMA trace. The rest of the paper is arranged as follows. Section II introduces a detachment mechanism that leaves no HMAs on climbing surfaces. Section III analyzes force and moment, and suggests how to improve the attachment performance with HMAs. Section IV presents our newly designed climbing robot. Section V shows the climbing experiment result of the robot. Section VI gives conclusions and future directions. II. T RACE -F REE D ETACHMENT M ECHANISM When using HMAs for robot climbing, attachment and detachment operations need to be carefully planned. In this section we propose a novel detachment mechanism to avoid the robot leaving an HMA trace on the climbing surfaces. The problem of HMAs being left on a climbing surface was caused before in a detachment operation. More specically, the climbing robot was designed with two feet and climbed in a rotational manner similar to that in Fig.

[nfabian, rebernet]@student.ethz.ch
978-1-4673-1405-3/12/$31.00 2012 IEEE

1181

Fig. 1. Trace-Free Detachment Mechanism for HMA-based robot climbing. The robot has two feet with HMAs on them. Within one climbing step, (a) the robot initially stays with both feet attached to a climbing surface (parallel to the paper). (b) It then detaches one foot (lower one) away from the surface without heating HMAs. (c) While rotating the body about the still-attached foot, the HMA on the detached foot is heated. (d) Once next foothold is reached, the foot attaches back onto the climbing surface.

Fig. 2. (a) A 0.97kg robot climbing on a vertical surface with 50 mm2 HMAs on each foot, as presented in the paper. (b) A free-body diagram of the robot on a vertical wall with a single foot attached with HMAs.

1 (more detail see [9]). However, different from Fig. 1b, detachment happened after the HMA on a foot had been heated up, so that the bonding strength could be lowered for easy operation. However that made the HMA become adhesive, according to the materials properties introduced in Section I. As a result, a trace was left behind, and new HMAs had to be supplied for subsequent steps. The detachment mechanism proposed in this paper, which we call the Trace-Free Detachment Mechanism (TFDM), primarily requires that a robots foot should only detach from a surface when HMAs are not adhesive. There is a temperature parameter called the bond formation temperature Tbf , below which HMAs are not adhesive [10]. When the robots foot detaches at a temperature Tdetach below Tbf , it is possible that no HMAs will be left on the surface. In the paper, we consider Tdetach at room temperature Tr (2530 C ) for control simplicity (Fig. 1b). TFDM also requires HMAs to remain on the robots foot, so that they can be used for multiple steps. That is achieved by exploiting HMAs bonding strength to different material. When HMAs are used to bond two surfaces, they form a bonding area A in between. The maximum bonding force Fmax when bonding breaks is proportional to A: Fmax = A (1)

Fig. 3. HMAs clamping bonding strength between material. TFDM requires a robots feet to be constructed with a material whose bonding strength with HMAs is higher than that of the climbing surface.

The proportion is HMAs bonding strength, which is a material dependent constant at a given temperature for a certain type of HMAs. In the case of HMA bonding between two different types of material, like a robots foot attached to a climbing surface (Fig. 2b), bonding will break at the surface where Fmax is smaller. By assuming the HMA bonding has the same area on both surfaces: Fmax = min(f oot A, wall A) (2)

For a climbing robot, two types of bonding forces need to be considered, these being a clamping bonding force F c which is normal to the surfaces, and a shear bonding force F s which is tangential to the surfaces (Fig. 2b).

To validate the relationship between bonding forces and bonding area, a set of experiments were carried out by manually breaking HMA bondings between two surfaces of the same material at Tr . The values of Fmax were recorded with a hand-held scale (VOLTCRAFT HS-10L), and HMA bonding areas A were measured after each trial. Experimental c result of the maximum clamping bonding force Fmax of HMAs to different material (i.e., copper and aluminium) c is shown in Fig. 3. Both Fmax and the maximum shear s bonding force Fmax of HMAs to a certain material (i.e., aluminium) is shown in Fig. 4 (more detail in Section III). Both gures conrm Eq. (1) that Fmax is propotional to A, and the proportion is material dependent (i.e., at Tr , c c Cu =0.98 MPa, Al =0.69 MPa). Table I further lists values of the clamping bonding strength c and the shear bonding strength s for a type of HMA to various material. The difference in the HMAs bonding strength to various materials is important for choosing the right constructing material for a robots feet. When the feet have a larger than the climbing surface, bonding will break at the climbing surface, and HMAs will remain on the feet: f oot >wall (3)

1182

TABLE I HMA S 1 B ONDING S TRENGTH 2 Material HMA Peltier element ceramic Stone Normal steel3 Anticorodal hard aluminium3 Copper ETP3 Roof batten r wood3 Window glass
1 2 3

WITH

VARIOUS M ATERIALS (MPa) Shear s (MPa) N/A 0.1 0.2 0.4-0.5 0.9-1.0 1.3-1.5 4.3-5.2 >2.0

Clamping 6.2-26.1 0.1-0.2 0.2-0.3 0.3-0.4 0.6-0.7 0.9-1.0 1.5-2.5 >2.0

Pattex Hot Sticks Transparent, Henkel, Germany Tdetach = Tr From ETH D-PHYS-Shop, https://lager.phys.ethz.ch/de/R/0200/ Fig. 4. Shear and clamping bonding forces determines an allowable range of HMA bonding area. When climbing on a surface, the lower boundary condition is determined by the robots weight. The upper boundary condition is determined by the motor force for foot detachment.

For example, when copper Electrolytic-Tough-Pitch (ETP) is used on the feet, a robot can climb without trace on stone, metallic and ceremic surfaces according to Table I. It is important to note that the cohesive strength of HMAs is higher than their bonding strength with other materials (Table I), which means bonding will only break at the climbing surface given Eq. (3), but not within HMAs. III. H ANDLING OF HMA S FOR ATTACHMENT This section considers how to handle the use of HMAs for attachment under TFDM. A piece of HMAs on the robots foot, on one hand has to provide enough force and moment for stable climbing, and on the other, has to be kept onboard for multiple climbing steps. With a set of force and moment analysis, we show how a minimal amount of HMAs can be decided for a stronger attachment. A. Shear and Clamping Bonding Force As Fig. 2b illustrates, the shear bonding force F s of HMAs by a single attached foot should be able to compensate the robots weight. In this case, the maximum shear bonding s of HMAs on the attached foot should be larger force Fmax than robots weight:
s Fmax > mg

Fig. 4 also shows an allowable range of HMA bonding area in the case of a robot climbing on a vertical surface: Amin < A < Amax (4)

And A should be as small as possible so that Fleg can be obtained from a motor with a smaller weight and torque. B. Shear Bonding Moment The shear bonding moment is important for stable climbing when the robot is moving. For example, the maximum s shear bonding moment Mmax for the attached foot should be larger than the moment of the robots weight, so that the robot does not fall off during rotation. Given the necessity of a minimal size of HMAs described in the previous s . For that purpose, we subsection, we try to augment its Mmax have developed a model of the shear bonding moment, which s suggests that multiple bondings generate a larger Mmax than a single bonding with the same total area. We assume the centre of gravity (COG) is in the middle w of a robot. The maximum moment of weight Mmax occurs when the robot is in a horizontal posture on a vertical surface, when the weight has the longest moment arm of half the s w robots length L. Mmax should be larger than Mmax :
s w Mmax > Mmax = mg

Considering Eq. (1), the smallest bonding area Amin shall be: mg Amin = s s From Table I, we can see is in the order of 0.1 MPa for most materials at Tr . That means, with as small as 1 cm2 bonding area on each foot, the HMA can keep a robot weighing more than 1 kg on most vertical surfaces. With TFDM, a minimal amount of HMAs is however preferred for the following reason. As Fig. 4 shows, while a s larger bonding area generates larger Fmax which enhances c stable climbing, it also leads to larger Fmax which requires a stronger motor for the detachment operation. Given a certain c motor force Fleg , Fmax is constrained by:
c Fmax < Fleg

L 2

(5)

Fig. 5a illustrates two cases of HMA bonding(s) on a single foot of a robot. A1 is a single-bonding case where the area is a circle with radius R1 , and A2 is a doublebonding case where the total area is equal to A1. The bonding forces in the two cases are the same according to Eq. (1). The maximum shear bonding moment M 1s max for A1 can be obtained by integral in a polar coordinate system:
2 s M 1s max = 0 0 R1

r2 dr d =

Therefore the largest HMA bonding area Amax should be Amax = Fleg c

2 s 3 R1 3

(6)

For A2, we assume the two bondings are identical and both located R0 from the centre of A1. Each can be seen as part
1183

Fig. 6. Hardware implementation of the climbing robot and one of its foot. TABLE II C OMPARISON OF T WO D ESIGNS OF HMA-BASED C LIMBING ROBOTS Spec Mass Dimension (LWH) Leg actuator torque Pedestal actuator torque Actuator number Peltier number Resistor number Speed HMA supply Trace Climbing surface Previous design [9] 0.60 kg 200120110 mm3 0.2 Nm (foot lift-up) 3 Nm 6 2 0 0.16 m/min Yes Yes Almost any New design 0.97 kg 260160120 mm3 3.06 Nm 2.90 Nm 4 4 8 0.45 m/min No No Not wood or glass

Fig. 5. (a) A model of maximum shear bonding moment for stronger attachment with HMAs. (b) With the same total area, a double-bonding case has larger moment than a single bonding. The model is validated by experimental data (star and cross marks).

IV. H ARDWARE I MPLEMENTATION of a ring with outer and inner radius R2 and R0 , and a span angle . The maximum shear bonding moment M 2s max is:
s M 2s max = 2 0 R0 R2

r2 dr d =

2 s (R2 3 R0 3 ) (7) 3

We assume each area can be approximated by a square: R0 = R2 R0 (8)

We have designed a new climbing robot with the above considerations for attachment and detachment operations. The robot is simpler than the previous version due to the absence of a module for supplying HMAs [9]. Each robots foot has been made larger to accommodate two pre-supplied distant HMA pieces based on the analysis in Section III. A. Overall Design The robot is 260 mm long and weighs 0.97 kg. It has two identical feet, and each foot has two actuators. One actuator is for rotating the body within the climbing surface, labelled as pedestal actuator in a blue lined circle in Fig. 6 (Kondo KRS-2350 HV ICS servomotor). It has enough torque to w rotate the robot on a vertical surface (Mmax = 1.30 Nm). The other actuator is for rotation away and towards the climbing surface, labelled as leg actuator in a green lined circle in Fig. 6 (Futaba S9157 servomotor). It can generate 100 N Fleg for detachment with a 30 mm lever arm. The robot is manually assembled with laser cut PMMA parts. More specications of the robot and a comparison to the previous version can be found in Table II. B. Foot Design As the most important part in the climbing robot, the foot requires careful design with three major principles. First, its surface material should have a higher than climbing surfaces in order to implement TFDM. Second, the foot must

Substituting Eq. (8) into Eq. (7) yields: 2 s (R2 R0 )(R2 3 R0 3 ) (9) M 2s max = 3R0 It is worth mentioning that the multiple-bonding method can be scaled up to any number of bondings in a similar way to Eq. (7) and (8). In this paper, we use a double-bonding case as it is sufcient to show the benet of multiple bondings. To validate the model, we conducted another set of expers iments, in which Mmax was measured for HMA bondings between aluminium and copper surfaces for both cases, with s two bonding areas 10 cm apart. Mmax was calculated as the product of the shear force when a bonding was broken manually, and a moment arm length of that shear force. Results from Fig. 5b shows that the model ts experimental data very well. With the same total area, the double-bonding s case results in a much larger Mmax , for example, 4 Nm 2 when the total area is 50 mm , while under 0.5 Nm with a single HMA bonding.

1184

Fig. 7. HMAs heating and cooling processes on the foot. Resistor heating and Peltier element cooling achieved 10 seconds and 22 seconds respectively between 30 C and 60 C .

Fig. 8.

Flow chart of an open-loop controller for robot climbing.

V. C LIMBING D EMONSTRATION AND E VALUATION be large enough to accommodate two distant pieces of HMAs for stronger attachment. Third, it has to reheat HMAs to a temperature higher than Tbf for (re)attachment. To fulll the rst requirement, we have selected copper ETP plate as the material to construct the foot surface (Fig. 6). From c listed in Table I, the robot is able to climb without trace on surfaces of aluminium, steel, ceramic, and stone, but not wood or glass. To choose the right foot size for the second requirement, we need to determine the two bonding areas and their distance. Take climbing a vertical aluminium surface as an example, the amount of the HMA on a foot should be within the range of 10 mm2 and 150 mm2 according to Eq. (4). Given the further constraint of Eq. (5), the minimal amount of the HMA becomes 25 mm2 , which requires two pieces 10 cm apart from (Fig. 5b). With a safety margine of 2.0, we used 50 mm2 of the HMA for each foot. The nal size of each foot is 16055 mm2 . For the third requirement, we use power resistors to heat and Peltier elements to cool the HMA. A relay switches between the two processes. Each foot uses four 5 power resistors linked in parallel and two 23W Peltier elements (TEC1-07105, 303040 mm3 , Centenary Materials), all connected to external power supplies. We have found that the speed of heating and cooling can be increased with higher power. As Fig. 7a shows, with a 45W power to the resistors, the heating time from 30 C to 60 C can be as short as 10 seconds. By using a heat sink, a fan (KDE1204PFVX, 1.8 W, Sunon), and separate copper plates (5030 mm2 ) for the two HMA pieces, the cooling time can be 22 seconds between 60 C and 30 C . Since the type of HMA in our study (Pattex Hot Sticks Transparent, Helkel) has a Tbf at 55 C , such a temperature range should be sufcient for (re)attachment. The design also has to consider the parallel alignment between foot and climbing surfaces, so that the HMA is in full contact with both. We use a copper plate for adjustment purpose (Fig. 6). In this section, we rst introduce an open-loop controller used by the robot to climb. We then demonstrate the robot climbing on a vertical surface without leaving HMA traces. Lastly, we evaluate the climbing speed improvement due to the design considerations for attachment and detachment. A. An Open-Loop Controller The four servomotors and the relay are controlled by in an open-loop manner as shown in Fig. 8. The target servo angles are pre-dened so that the robot climbs up straight. The motor commands and relay control signals are based on the heating and cooling time as explained in Section IV. The controller is implemented in Matlab and communicated by an Arduino Duemilanove micro-controller on the robot. B. Climbing Performance Climbing experiments were carried out on an indoor vertical smooth aluminium surface. In a total of 15 trials, the robot succeeded in climbing multiple steps (maximum 4) before falling off. In all trials, the robot detached its feet for a sum of 38 times, no HMA trace was ever left on the surface. Fig. 9 shows a series of important snapshots of the robot in a two-step climbing (also see supplementary video le). Fig. 10 shows the trajectories of the COG and both feet during the two steps. The height data was extracted from the same trial as shown in the video. Note that the heating time was around 25 seconds in the trial, which is longer than the 10-second resistor heating time in Section IV.B. That is because the servos speeds were set low for smooth motions, which include a foot lift-up during detachment, a rotation for inclining, and a foot push-down before attachment. The energy efciency was also calculated as the useful work that contributed to overcome the robots weight, divided by the electrical energy cost of heating and cooling in a single step. The value was below 0.1%. The energy efciency was not high for continuous climbing, however it is worth mentioning that when the robot needs to stop on a vertical surface for a long period, no additional energy will be needed due to the long term effect of HMA bondings.

1185

VI. C ONCLUSIONS AND F UTURE W ORKS The HMA-based approach has a great potential use for climbing applications, due to its high attachment strength, repeatable use on different surfaces, and ease of implementation. This paper shows design considerations for attachment and detachment operations to improve the climbing performance of this approach. We have proposed a novel detachment mechanism (TFDM) which exploits the adhesive property of HMAs and their bonding strength to different material. By having the feet constructed from copper material, the robot can climb on a variety of smooth vertical surfaces, as long as their clamping bonding strength with the HMA is below 1 MPa at room temperature. The feet must detach from the climbing surfaces when the HMA is below 55 C in order to climb without leaving a trace. We have also developed a set of feasible models to dene the allowable amount range of HMAs for attachment use. Result shows that with multiple bonding areas, the attachment can be made stronger with a given minimal amount of HMAs. Through a newly designed robot, we have demonstrated multiple step climbing at 0.45 m/min on a vertical surface. In the future, we will continue to work towards more climbing steps, which will require a more compliant adjustment mechanism to guarantee parallel alignment between the feet and the climbing surfaces. We are also interested in making the robot fully autonomous, meaning that the control programme and a battery will be onboard. R EFERENCES
[1] M.P. Murphy and M. Sitti, Waalbot: An Agile Small-Scale Wallclimbing Robot Utilizing Dry Elastomer Adhesives, IEEE-ASME T. Mech., vol. 12, 2007, pp. 330-338. [2] H. Prahlad, B. Pelrine, S. Stanford, J. Marlow, and R. Kornbluh, Electroadhesive Robots - Wall Climbing Robots Enabled by a Novel, Robust, and Electrically Controllable Adhesion Technology, in 2008 International Conference of Robotics and Automation, Pasadena, CA, USA, 19-23 May, 2008, pp. 3028-3033. [3] S. Kim, M. Spenko, S. Trujillo, B. Heyneman, D. Santos, and M.R. Cutkosky, Smooth Vertical Surface Climbing with Directional Adhesion, IEEE T. Rob., vol. 24, 2008, pp. 65-74. [4] K.A. Daltorio, T.E. Wei, A.D. Horchler, L. Southard, G.D. Wile, R.D. Quinn, S.N. Gorb, and R.E. Ritzmann, Mini-Whegs TM Climbs Steep Surfaces using Insect-Inspired Attachment Mechanisms, Int. J. Rob. Res., vol. 28, 2009, pp. 285-302. [5] M. Murphy, C. Kute, Y. Menguc, and M. Sitti, Waalbot II: Adhesion Recovery and Improved Performance of a Climbing Robot using Fibrillar Adhesives, Int. J. Rob. Res., vol. 30, 2011, pp. 118-133. [6] B.L. Luk, A.A. Collie, and J. Billingsley, Robug II: An Intelligent Wall Climbing Robot, in 1991 International Conference of Robotics and Automation, Sacramento, CA, USA, 7-12 April, 1991, pp. 23422349. [7] Y. Wang, S. Liu, D. Xu, Y. Zhao, H. Shao, and X. Gao, Development and Application of Wall-Climbing Robots, in 1999 International Conference of Robotics and Automation, Detroit, MI, USA, 10-15 May, 1999, pp. 1207-1212. [8] H. Amano, K. Osuka, and T.-J. Tarn, Development of Vertically Moving Robot with Gripping Handrails for Fire Fighting, in 2001 IEEE/RSJ International Conference on Intelligent Robots and Systems, Maui, HI, USA, 29 October-3 November, 2001, pp. 661-667. [9] M. Osswald and F. Iida, A Climbing Robot based on Hot Melt Adhesion, in 2011 IEEE/RSJ International Conference on Intelligent Robots and Systems, San Francisco, CA, USA, 25-30 September, 2011, pp. 5107-5112. [10] G. Gierenz and W. Karmann, Adhesives and Adhesive Tapes, WileyVCH, Weinheim, Germany; 2001.

Fig. 9. Snapshots of climbing two steps on a vertical surface. (a) Initial position: both feet are attached to the surface with HMAs. (b) Lower foot is detached. (c) First step by rotating around the upper foot by 180 . The HMA on the detached foot is heated during the rotation. (d) When the target position is reached, detached foot is attached back to the surface. (e) Peltier elements cool down the HMA. (f)-(i) Repeat (b)-(e) for a second step.

Fig. 10. Trajectories of robots two feet and COG. Labelling is consistent with Fig. 9. The initial height of COG is the reference height.

C. Speed Evaluation Speed is important for climbing locomotion, but was not satisfactory in our previous robot. That was partly due to slower heating and cooling processes on the feet (70 seconds in one step [9]), which has been improved in the new robot to 10 seconds for heating and 22 seconds for cooling. The other reason was the presence of an HMA supply process (11 seconds in one step [9]) for feeding new HMAs for new steps. With TFDM proposed in this paper, the same piece of HMA was (re)used for multiple steps and the HMA supply process was eliminated. As a result, the new robot climbed at 0.45 m/min, which is 3 times as fast as the previous version.

1186

You might also like