You are on page 1of 6

2012 IEEE/RSJ International Conference on Intelligent Robots and Systems October 7-12, 2012.

Vilamoura, Algarve, Portugal

Climbing Vertical Terrains with a Self-Contained Robot


Liyu Wang, Lina Graber and Fumiya Iida

Abstract Vertical climbing on a variety of at surfaces with a single robot has been previously demonstrated using vacuum suction, electrostatic adhesion, and biologically inspired approaches, etc. These methods generally have a low attachment strength, and it is not clear whether they can provide satisfactory attachment on vertical terrains with richer 3D features. Recent development of a climbing technology based on hot melt adhesives (HMAs) has shown its advantage with a high attachment strength through thermal bonding and viability to any solid surfaces. However, its feasibility for vertical climbing has only been proven on at surfaces and with external energy supplies. This paper provides quantitative measurements for vertical climbing performance on ve types of surfaces and terrains with a self-contained robot exploiting HMAs. We show that robust vertical climbing on multiple terrains can be achieved with reliable high-strength attachment.

I. INTRODUCTION Robotic climbing has been a popular research topic since the invention of a stair-climbing robot [1] in 1960 and the rst demonstration of vertical wall climbing over 25 years ago [2]. It was estimated over 200 prototypes of climbing robots had been developed by 2005 [3] with many more appearing since. Climbing robots have useful applications such as inspection, structure testing, civil construction, cleaning, vertical transportation, and security, etc [4]. Several approaches have been suggested to generate counter-gravity forces for vertical robotic climbing (see recent reviews [4], [5], [6] with further earlier reviews referred in [5]). Among them, some have shown the capability of a single robot climbing on various vertical at surfaces, by using e.g. passive or active vacuum suction [7], biologically inspired methods with dry adhesives [8], [9], directional microspine arrays [10], [11] or micro-structured polymer [12], as well as electrostatic adhesion [13], etc. The attachment strength of those methods was generally low (103 104 Pa), which means a large contact area is mandatory to achieve a high payload. These approaches have also not shown whether they could provide satisfactory attachment on vertical terrains that have richer 3D features than at surfaces. Methods with higher attachment strength exists by means of magnetic engagement (105 Pa), however its use is restricted to ferromagnetic surfaces. Hence, the challenge still remains to design climbing robots with a high attachment strength for multiple vertical surfaces and terrains.
*This research was supported by the Swiss National Science Foundation Professorship Grant No. PP00P2123387/1, and the ETH Zurich Research Grant ETH-23-10-3. L. Wang, L. Graber, and F. Iida are with Department of Mechanical and Process Engineering, ETH Zurich, 8092 Zurich, Switzerland

Recently, a novel climbing technology was developed using unconventional material hot melt adhesives (HMAs) [14], [15], [16]. The main advantage is the high attachment strength resulting from thermal bonding (104 -106 Pa depending on the type of HMA and material of surfaces [15], [16]): Theoretically, only an area of 4 cm2 HMA is needed for a robot to hold a combination of its own mass and payload up to 60 kg on a vertical surface [14] (subject to mechanical strength of the robot). Furthermore, HMAs can be applied to almost any solid surface, and successful multi-step climbing has been demonstrated on vertical at surfaces of wood [14] and metal [16]. However, due to the presence of thermal processes, all previous robots were powered by external energy sources, but the energy cost of this method has not been measured. The potential advantage from applying HMAs on vertical terrains has not yet been fully explored, therefore the climbing performance of a robot on environments such as curved surfaces, stairs and uneven terrains, etc., remains unclear. The main contribution of the paper is to present quantitative measurements of climbing on multiple vertical terrains with a self-contained robot exploiting HMAs. We rst introduce a newly prototyped climbing robot with onboard batteries and controller. We then demonstrate multi-step climbing with the self-contained robot on a vertical at surface, and detail the energy cost of the approach. We nally assess its climbing performance on multiple vertical surfaces and terrains, and show that success rates of attachment and detachment during climbing are high under all conditions tested. The remainder of the paper is arranged as follows: Section II gives a brief summary of the climbing technology based on HMAs. Section III introduces the self-contained climbing robot. Section IV shows the energy consumption of the robot in multi-step climbing demonstrations on a vertical at surface. Section V presents quantitative measurements of attachment and detachment during robot climbing on multiple surfaces and terrains. Section VI concludes the paper and points out possible applications of the robot. II. CLIMBING BASED ON HOT MELT ADHESIVES In this section, we briey introduce attachment, detachment and locomotion in HMA-based climbing. As a mixture of polymer and other ingredients, HMAs not only exhibit temperature-related rheological properties but also temperature-dependent adhesive bonding properties [17]. The material is solid at room temperature with storage modulus and tensile strength both over 10 MPa. It generally softens at 60 C-80 C, and melts into liquid above 150 C. The phase
305

[liyu.wang, fumiya.iida] at mavt.ethz.ch, graberl at student.ethz.ch


978-1-4673-1735-1/12/S31.00 2012 IEEE

transition is bi-direction and repeatable. Here we assume that the climbing environment is at room temperature. A. Attachment through Thermal Bonding Although it is possible to directly exploit HMAs adhesiveness at higher temperatures for attachment [18], a cooling process with heated HMAs can provide much higher attachment strength. This thermal bonding characteristic has been studied in polymer science and can be explained by the theory of wetting. Thermal bonding requires a climbing robot to attach itself onto a surface with heated HMAs. HMAs may come from an onboard supplier [14], [15], or may be preloaded and kept on a robots feet [16]. A bonding area can be formed between the robot and the surface during cooling, which can be either passive (as in [15] and this paper), or active with Peltier elements or fans [14], [16]. Attachment performance may be further improved with multiple bonding areas, i.e. several HMA pieces on a robots foot with a distance between each other. This not only increases the shear bonding moment [16], but also offers higher possibility for bond formation, which will be assessed in this paper on multiple terrains. B. Detachment Detachment can be achieved in two ways, by heating up the HMA bond, or by breaking the bond at room temperature. The former way requires longer time for heating up, but a small motor force would be enough for detachment [14]. The latter way can save the heating time, but it requires a relatively large motor force [16] or a torsion [15]. Due to the adhesiveness of HMAs at higher temperatures, the former way is highly likely to result in HMA remains on climbing surfaces [14], which would restrict the application of the technology. This problem has been solved by exploiting the difference in bonding properties of material between the robots feet and climbing surfaces [16]. C. Locomotion So far, all prototyped climbing robots based on thermal bonding with HMAs have adopted biped legged locomotion. Two legs of the robots take turns to attach and detach feet, with a period of double-foot attachment in between. The robots climb by rotating their bodies about the attached foot. By controlling the rotational angle, a robot can move towards any direction. Because there are distinctive processes of heating, cooling, attachment and detachment, a legged locomotion pattern is more intuitive for controllers design. However, due to the need for rotation, it requires high-torque motors when larger mass+payload shall be lifted up. Other types of locomotion such as using wheels should also be possible though not yet explored. III. A SELF-CONTAINED CLIMBING ROBOT We have extended the HMA-based climbing technology with a newly prototyped robot (Fig. 1a). The latest robot features a more efcient heating method, self-contained energy supply and control, and four pads on each foot for
306

3 batteries pedestal actuator battery carrier pedestal actuator

leg actuator

(a)

9 batteries

(b)

Fig. 1. A self-contained climbing robot based on thermal bonding with HMAs. (a) Overview of the robot with 12 batteries mounted in two groups (3+9) on a battery carrier cover. (b) Mechanical design of the robot with three actuators. TABLE I S PECIFICATIONS OF THE S ELF -C ONTAINED C LIMBING ROBOT Mass (including batteries) 1.25 kg Dimension (LWH) 246246140 mm3 11511510 mm3 Foot dimension (LWH) 2015 mm2 Pad area (LW) Neighbouring pad distance 95 mm Leg actuator torque 2.90 Nm 2.90 Nm Pedestal actuators torque 9.9 V Nominal voltage Estimated energy storage 36 Wh Energy consumption (static) 0 Wh 0.88 Wh Energy consumption (per step)

multiple HMA bonds. The robot carries a circular cover where batteries are mounted. This battery carrier determines the robot diameter of 246 mm. The robot has two legs made from aluminium bars, whose lengths are 108 mm. They can be rotated with respect to each other with a servomotor i.e. leg actuator in Fig. 1b (KRS-2350 HV, Kondo, Japan), so that one of the legs can move towards or away from climbing surfaces. At the end of each leg, there is a foot which can be rotated by another servomotor i.e. pedestal actuator in Fig. 1b (KRS-2350 HV, Kondo, Japan). That enables the robot to generate a torsion for detachment, as well as to rotate its body within the climbing surface in any target direction. The robot is manually assembled. More specications can be found in Table I. A. Foot Design As the most important part of the climbing robot, the foot requires careful design for reliable climbing. Its primary function is to heat and cool HMAs between the foot and a climbing surface for attachment and detachment. This happens on a copper pad, because copper does not easily deform and has a high heat conductivity. Heating was previously realized with power resistors. It required 10 seconds to heat

Fig. 2.

Design of the foot and the pad. Fig. 3. Flowchart of an open-loop controller implemented onboard.

a copper pad (5030 mm2 ) from room temperature to 60 C with a 45 W power [16]. To make heating more efcient, we have replaced power resistors with a resistance heating wire (Kanthal A, 14.45 /m, Sandivik, Sweden), so that the wire (10 cm) can be winded and closely attached to the back of a copper pad (2015 mm2 ) and heat can be transferred faster (Fig. 2). We used a second copper piece of the same size to support the resistance wire so that it can be tightly located between the two copper pieces (Fig. 2). The sandwich-like structure can be fastened by screws onto the foot, which is made of wood with surfaces coated by a heat-resistant silicon-based paint (Supertherm, MOTIP DUPLI, Germany) (Fig. 2). In this way, heat and electricity can be both insulated on the foot, and it is strong enough to support the robot at a high temperature. In 10-second heating time, temperature of a pad can be increased from room temperature to 90 C with a 50 W power (measured by a multimeter [UT61B, UNITREND, China]). Cooling is achieved passively for a battery saving purpose, and it could take 5-6 minutes from 90 C to room temperature. Another function of the foot is to provide stable attachment while being compliant. It has been modelled that with the same total area of HMA bonding, multiple distributed bonds can generate much higher shear attachment moment than a single bond [16]. This principle is continued in the current design, where four pads are used on each foot so that the probability of attachment can also be increased (Fig. 2). The distances between the pads are 95 mm, resulting a total foot surface of 115115 mm2 . B. Batteries and Their Carrier Onboard energy supply has been realized with the present robot. Considering the potential high demand for heating, 12 rechargeable lithium iron phosphate batteries (APR 18650, A123 Systems, USA) are used. They are connected in a 4p3s conguration, so that 9.9 V can be supplied to the servomotors and a microcontroller (further detail in next subsection). Since each battery has a nominal voltage of 3.3 V and a nominal electric capacity of 1.1 Ah, a total energy storage of 36 Wh can be obtained if batteries are fully charged (assuming 90% discharge). However, batteries result in 0.5 kg additional mass of the robot. We use a mass307

shifting battery carrier cover to reduce the needed torque on the pedestal actuators. The light-weighted carrier is made from a commercial microwave plate, and batteries are glued along its edge in two groups (3+9) opposite to each other. During vertical climbing, the cover will passively position itself so that the group with more (nine) batteries will always stay at the lowest point of the robot body (Fig. 1a). C. Electronics The physical level control consists of a microcontroller board (Duemilanove, Arduino, Italy) and two relays (G5LE1-VD, Omron, Japan). Different from the previous prototypes where the microcontroller received control commands from a desktop computer, this robot performs all its actions with the microcontroller. The onboard programme controls the three servomotors and the timing of active heating and passive cooling processes. For each foot there is a relay for switching between heating and cooling: Depending on the signal from the microcontroller, the resistance wires are either connected to the batteries or disconnected. Resistance wires inside the four pads on each foot are connected in a 2p2s conguration, resulting in a resistance of 1.5 . IV. CLIMBING ON VERTICAL FLAT SURFACES As reviewed in Section II, HMA supply issue has been solved either by a supplier or by reusing preloaded HMAs on the feet, we assume in this paper that sufcient HMA (Pattex Hot Stick Transparent, Henkel, Germany) are always available on pre-coated climbing surfaces (and terrains in the next section). We focus on demonstrating the feasibility of vertical climbing with two challenges imposed by selfcontainment: 1) The robot has to deal with extra weight resulted from onboard batteries; 2) The number of steps is constrained by the energy storage of the batteries. In this section, we rst show self-contained multi-step climbing on a vertical at surface with an open-loop controller. We then show the energy cost in a two-step climbing case study. A. Locomotion Control As briey introduced in Section II, climbing locomotion can be controlled by sequential motor commands and switch-

TABLE II E NERGY C OST IN THE C LIMBING ROBOT BASED ON HMA S Action Heat foot 1 Detach foot 1 Move body Attach foot 1 Cool foot 1 (passive) Heat foot 2 Detach foot 2 Move body Attach foot 2 Cool foot 2 (passive) 2-step climbing Time cost (s) 15 4 10 12 340 15 4 10 12 340 762 Energy cost (Wh) 0.21 0.06 0.14 0.17 0.19 0.20 0.06 0.14 0.17 0.46 1.77

Fig. 4. Snapshots of climbing two steps on a vertical at surface. (a) Initial position: both feet are attached to the surface coated with HMAs. (b) Lower foot (denoted as 1) is detached with torsion from pedestal actuator 1. (c) First step climbing by rotating about foot 2. (d) When the target position is reached, foot 1 attaches to the surface and heats HMAs. (e) Passively cooling down HMAs on the surface. (f)-(j) Repeat (a)-(e) for the second step. Labelling corresponds to processes in Fig. 3.

ing between heating/cooling processes. In the self-contained robot, the control programme is preloaded onto an onboard microcontroller (see Section III.C). The programme can be described as an open-loop owchart shown in Fig. 3. It is worth mentioning that we used the combination of the two detachment methods (as introduced in Section II.B) to increase the reliability. The target heating and cooling time is based on the empirical speed of active heating and passive cooling on the pad (refer to Section III.A) by adding safety margins considering the heat conductivity and the specic heat capacity of HMAs. The target servomotor angles for the leg actuator are predened so that both feet can be lifted up (40 ) from climbing surfaces after detachment and then back on for attachment. The target servomotor angles for the pedestal actuators are also predened so that the robot can not only generate enough torsion (with a 45 angular displacement) for foot detachment, but also climb up straight. As a result, the robot managed to demonstrate repeatable multiple (5-12) steps climbing on vertical at surfaces of a wood block board and a coated chip board. Fig. 4 shows snapshots of a two-step climbing on a at coated chip board surface, where each snapshot corresponds to a process in the owchart in Fig. 3 (also see Part 1 in a supplementary video). The upper ve snapshots show the climbing step of foot 1, while the lower ve show that of foot 2. The maximum possible number of steps (regardless of mechanical failures), however, is constrained by the energy stored in the batteries, and determined by the energy cost per step, which we will show in the following subsection. B. Energy Consumption Energy consumption is important for self-contained climbing, but little has been done before to analyze the en308

ergy cost in the HMA-based climbing technology. It was only briey mentioned that the electrical-potential energy conversion efciency was below 0.1% [16]. We therefore have detailed the energy cost per step of the robot. For the purpose of recording power, we used a 9.9 V voltage from an external power supply (VOLTCRAFT VLP-1405 PRO, Conrad, Switzerland) which can display the real-time current. Table II shows the corresponding energy cost for each process in the control owchart (Fig. 3). Here we can see the heating and cooling time lengths, and that passive cooling takes 89.2% of the time cost for a single step. As explained in Section III.A, that is for a battery saving reason, from which we see a limitation of the technology i.e. a trade-off between energy cost and time cost. We can also notice that the energy cost during passive cooling was not zero and it differed largely between steps. That can be explained by the rising of current through the leg actuator as it continuously exerted a press force onto the climbing surface. This energy cost can be reduced by turning the leg actuator off shortly after the start of cooling. V. PERFORMANCE ON VERTICAL TERRAINS Terrains are more than surfaces as they contain 3D features. While robot locomotion on land terrains has been extensively studied, there has been few study on climbing vertical terrains to the best of our knowledge (roughness was discussed with vertical at surfaces in [10]). Commonly seen terrains for climbing robots include at or curved surfaces, stairs, and uneven terrains, etc. All these terrains require a robot to accommodate their contours (if not with additional degrees-of-freedom) so that contact areas can be maximized to provide sufcient attachment force. In this section, we assess the climbing performance of the robot on ve types of terrains placed vertically, i.e. a at surface, a curved convex surface, a stair, and two uneven terrains. We quantify each terrain with a metric called Terrain Shape Index (TSI) [19], and measured HMA bonding area, number of attached pads, and success rate for making a single climbing step. We show that the success rates are high for all ve types of terrains. A. Terrain Shape Index It was proposed in geography that terrains can be quantitatively described by TSI [19], which is a metric representing the ratio of the mean height difference between a reference

Fig. 6. Boxplot results showing attachment on multiple vertical terrains. (a) HMA bonding area. (b) Number of pads attached. TABLE III S UCCESS R ATE V. S . T ERRAIN S HAPE I NDEX Type of vertical surface/terrain Flat surface of a coated chip board (F) Curved convex surface of plastics (C) Stair of aluminium alloy (S) Uneven terrain of stones (UTS) Uneven terrain of wood plates (UTW) Fig. 5. The robot on four types of vertical terrains including (a) a convex surface C, (b) a stair S, (c) an uneven terrain of stones UTS, and (d) an uneven terrain of wood plates UTW. TSI 0 -0.114 0.167 -0.005 0.038 Success rate 100 % 83 % 100 % 83 % 80 %

point and its surroundings to the radius of the surroundings. TSI is more than just a surface roughness measurement (e.g. in [10], [20]), since it considers both horizontal and vertical dimensions. It is a value which depends on the reference point, the sampling points in its surroundings, and sampling plot radius. Assuming a robot is at a certain location (which is the reference point), and the height there is noted as Z0 , we select N sampling points around that location with a sampling plot radius R. The mean difference in height Z between the reference point and its surroundings is:
N

(1) N where Zi is the height of the ith sampling point. The sign of Z will be positive for a concave topology and negative for a convex topology. A value of zero indicates a at smooth surface. TSI can be then dened as : T SI = B. Vertical Terrains Five types of vertical terrains were prepared for the performance assessment. The rst type was the same at surface of a coated chip board (as in Section IV, denoted as F). The second terrain was a curved convex surface, which we used the outer surface of a plastic cylindrical container
309

Z=

i=1

(Zi Z0 )

with a mean diameter of 32 cm (Fig. 5a, denoted as C). The third type was a single stair, with the lower level on a at coated chip board and the higher level on an aluminium alloy block with a height of 5 cm (Fig. 5b, denoted as S). The fourth was an uneven terrain by gluing irregular stones together on a at coated chip board surface (Fig. 5c, denoted as UTS). And the fth was an uneven terrain by randomly overlying several wood plates (thickness 5 mm) on a at coated chip board surface (Fig. 5d, denoted as UTW). All ve terrains were pre-coated with HMAs for the reason explained at the beginning of Section IV. For each terrain, TSI was calculated by selecting a reference point and eight sampling points equally (3 cm) located to each other in a half circle of 15 cm radius (1.5 times the leg length) towards the direction of climbing (detailed values in Table III). The reference points were used as the initial position for the centre of the robot. Note that the reference point in the case of the stair was located on the lower level 12 cm from the edge of the stair, so that the robot can reach the higher level within a single step. C. Performance Assessment Attachment and detachment in climbing was assessed on the above vertical terrains with 10-12 single-step trials in the cases of F, C, UTS, and UTW, and seven 2-step trials in the case of S (also see Part 2 in a supplementary video for cases C, S, UTS, and UTW). Three parameters were measured including HMA bonding area, number of attached pads, and success rate. Here success was dened as the robot being able to attach one foot, to detach the other foot and nally

Z R

(2)

to hold itself on the vertical terrains as a complete climbing step. Fig. 6 shows boxplots of HMA bonding area and number of attached pads on all the terrains. Plus signs (+) indicate outliers which are numerically distant from the rest of the results and should not be considered. Medians are indicated by a horizontal thick line in the boxes, while lower and higher quartiles are indicated by bottom and top boundaries of the boxes. The median values for HMA bonding area were 6.0 cm2 , 3.0 cm2 , 4.5 cm2 , 5.6 cm2 and 2.3 cm2 for cases F, C, S, UTS and UTW respectively. More than 75% of the trials on all the terrains had an HMA bonding area of over 2.0 cm2 , which is sufcient to hold the weight of the robot. The median values for number of attached pads are 2, 1.5, 2, 3 and 1.5 out of 4 pads correspondingly, with more than 75% of the trials on all the terrains having at least one pad successfully attached. Table III shows overall success rates with respect to TSI. It can be seen that although TSI varied between positive and negative values in a difference of two orders, success rates are always high (80%-100%) using thermal bonding with HMAs. The above results can be explained on one hand by the fact that distributed pads can increase the probability of HMA bond formation, as indicated the above result on number of pads attached. On the other hand, HMAs rheological property at higher temperature makes the robot more accommodating towards multiple vertical terrains with a difference in height (or depth) one or two orders below the robots dimension. Possible limitation comes from the leg actuator servomotor range and distance between pads, as they determines the maximum depth difference the robot can overcome and the minimum diameter of potential obstacles (e.g. a high sharp pole). VI. CONCLUSIONS This paper presents a self-contained robot and an assessment of its climbing performance on multiple vertical surfaces and terrains. The main advantage of this approach results from a thermal bonding effect of unconventional material HMAs. Multi-step climbing has been demonstrated on a vertical at surface with onboard batteries storing 36 Wh energy, and the energy consumption of the robot was measured as 0.88 Wh per step. Experimental results suggest that reliable climbing with a success rate of 80%-100% can be achieved on a variety of vertical surfaces and terrains, whose Terrain Shape Indices range from -0.114 to 0.167. Despite the advantage of high strength and reliable performance on multiple vertical surfaces and terrains, the approach has a trade-off between time cost and energy cost. This suggests that self-contained climbing with HMAbased technology favours applications where speed is not the priority but payload is more important, such as vertical transportation on rough terrains. Alternatively, due to the long-term bonding effect of HMAs, the robot may be used in tasks where the static period is much longer than its movement period, such as designated patrolling for security purposes.
310

R EFERENCES
[1] C. A. King, Occupant-controlled, self-propelled, obstruction-climbing vehicle, U.S. Patent 2 931 449, Apr. 5, 1960. [2] A. Nishi, Y. Wakasuji, and K. Watanabe, Design of a robot capable of moving on a vertical wall, Adv. Robot., vol. 1, no. 1, pp. 33-45, Jan. 1986. [3] S. Chen, J. Shang, Z. Zhao, T. Sattar, and B. Bridge, Novel solutions to design problems of industrial climbing robots, in Proc. 8th Int. Conf. Climbing and Walking Robots and the Support Technologies for Mobile Machines, London, 2005, pp. 139-146. [4] M. F. Silva and J. A. Tenreiro Machado, A survey of technologies and applications for climbing robots locomotion and adhesion, in Climbing and Walking Robots, B. Miripour, Ed. Rijeka: InTech, 2010, pp. 1-22. [5] D. Longo and G. Muscato, Adhesion techniques for climbing robots: State of the art and experimental considerations, in Proc. 11th Int. Conf. Climbing and Walking Robots and the Support Technologies for Mobile Machines, Coimbra, 2008, pp. 6-28. [6] B. Chu, K. Jung, C.-S. Han, and D. Hong, A survey of climbing robots: Locomotion and adhesion, Int. J. Precis. Eng. Manuf., vol. 11, no. 4, pp. 633-647, Aug. 2010. [7] S. Hirose, A. Nagakubo, and R. Toyama, Machine that can walk and climb on oors, walls and ceilings, in Proc. 5th Int. Conf. Advanced Robotics, Pisa, 1991, pp. 753-758. [8] C. Menon and M. Sitti, A biomimetic climbing robot based on the gecko, J. Bionic Eng., vol. 3, no. 3, pp. 115-125, Sep. 2006. [9] M. P. Murphy and M. Sitti, Waalbot: An agile small-scale wallclimbing robot utilizing dry elastomer adhesives, IEEE-ASME Trans. Mechatron., vol. 12, no. 3, pp. 330-338, Jun. 2007. [10] A. T. Asbeck, S. Kim, M. R. Cutkosky, W. R. Provancher, and M. Lanzetta, Scaling hard vertical surfaces with compliant microspine arrays, Int. J. Robot. Res., vol. 25, no. 12, pp. 1165-1179, Dec. 2006. [11] M. J. Spenko, G. C. Haynes, J. A. Saunders, M. R. Cutkosky, A. A. Rizzi, R. J. Full, and D. E. Koditschek, Biologically inspired climbing with a hexapedal robot, J. Field Robot., vol. 25, no. 4-5, pp. 223-242, Apr.-May 2008. [12] 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. Robot. Res., vol. 28, no. 2, pp. 285-302, Feb. 2009. [13] H. Prahlad, R. Pelrine, S. Stanford, J. Marlow, and R. Kornbluh, Electroadhesive robots wall climbing robots enabled by a novel, robust, and electrically controllable adhesion technology, in Proc. 2008 IEEE Int. Conf. Robotics and Automation, Pasadena, pp. 30283033. [14] M. Osswald and F. Iida, A climbing robot based on hot melt adhesion, in Proc. 2011 IEEE/RSJ Int. Conf. Robotics and Intelligent Systems, San Francisco, pp. 5107-5112. [15] F. Rochat, V. Hirschmann, T. Barras, H. Bleuer, and F. Mondada, Climbing robot with thermal glue, in Proc. 14th Int. Conf. Climbing and Walking Robots and the Support Technologies for Mobile Machines, Paris, 2011, pp. 409-414. [16] L. Wang, F. Neuschaefer, R. Bernet, and F. Iida, Design considerations for attachment and detachment in robot climbing with hot melt adhesives, in Proc. 2012 IEEE Int. Conf. Robotics and Automation, Saint Paul, pp. 2792-2797. [17] W. Li, L. Bouzidi, and S. S. Narine, Current research and development status and prospect of hot-melt adhesives: a review, Ind. Eng. Chem. Res., vol. 47, no. 20, pp. 7524-7532, Sep. 2008. [18] E. Guizzo, 4 ways to climb a wall, robot edition (video), IEEE Spectrums Blogs of Automation, Jan. 18, 2010. URL: http://spectrum.ieee.org/automaton/robotics/industrial-robots/4-waysto-climb-a-wall-robot-edition-video, accessed Mar. 4, 2012. [19] W. H. McNab, Terrain Shape Index: Quantifying effect of minor landforms on tree height, Forest Sci., vol. 35, no. 1, pp. 91-104, Mar. 1989. [20] J. T. Black and R. A. Kohser, DeGarmos Materials and Processes in Manufacturing (11th Ed.). USA: John Wiley & Sons, 2011, pp. 1032-1034.

You might also like