Design and development of a sealing-system for climbing-robots

Transcription

1 Design and development of a sealing-system for climbing-robots T. LEICHNER, T. GASTAUER and B. SAUER Machine Elements, Gears and Transmission, Department of Mechanical Engineering, University of Kaiserslautern, Kaiserslautern, Germany C. HILLENBRAND, D. SCHMIDT and K. BERNS Robotics Research Lab, Department of Computer Science, University of Kaiserslautern, Kaiserslautern, Germany fi cahillen@informatik.uni-kl.de ABSTRACT: The Institute of Machine Elements, Gears and Transmissions (MEGT) cooperates with a Robotics Research laboratory at the Department of computer science. This research group develops a climbing robot, which will be able to survey concrete walls in order to detect irregularities and errors in the structure of the wall. As the final result the robot should be able to drive without any external support, fixed at the wall by using a regulated system of seven vacuum chambers. The climbing-robot will not only be able to stick to the wall but will also be able to move on it while scanning the entire surface. These two requirements of fixing the body to the wall and simultaneously moving on it, demands a very tricky sealing-system, which will be developed at the institute of MEGT. In order to guarantee that the climbingrobot keeps contact to the wall, it is necessary to develop a special sealing which needs to be wear-resistant, leak-proof and easily sliding. By now several prototypes have been developed in order to test different sealing materials and shapes in order to find the optimum in friction and wear. Keywords: climbing robot, friction, wear, pin on disc, sealing 1. INTRODUCTION The non-destructive inspection of large concrete walls is still unsolved. One option is to use a vacuum system for the adhesion and driven wheels for the propulsion. The seals for the vacuum chambers are slipping over the rough surface. Therefore it cannot be guaranteed that the chambers are always airproof. Especially regarding concrete walls a special seal design has to be found to increase the security of the adhesion. On the other side the propulsion system must be able to produce enough force for carrying and accelerating the robot to a suitable velocity. This paper will present the climbing robot Cromsci which uses the already mentioned techniques. The propulsion system consists of three omnidirectional driven wheels which are airproof and completely rotatable and it has been presented in earlier papers. For adhesion a vacuum system of seven controllable vacuum chambers and one reservoir chamber is used. This system including chambers and seals will be discussed in detail afterwards. The requirements for the sealing are very high concerning leak tightness and attrition due to the 1 / 10

2 rough and sharp-edged surfaces of concrete walls. Therefore, each sealing must be flexible enough to allow a good adaption to the ground but at the same time let the robot slide when the wheels are turning. Regular inspections and maintenance of concrete buildings like motorway bridges and dams are very time and cost expensive. In addition to that the technical staff has to operate in risky situations, e.g. by using complex access devices. For a better objectivity and reproducibility of inspections as well as for safe working conditions a climbing robot has been developed which can check the building remote controlled. Many experiments with the prototypes underlined the importance of the sealing: Although promising results were obtained on smooth concrete or wooden surfaces, the sealing failed on rough surfaces, as they occur on every concrete building. Therefore new sealing materials and designs were developed, as the existing adhesion concept turned out to be inadequate. In literature other climbing robots using negative pressure adhesion are presented, although not all of them can be used for this application [1] [2] [3] [4]. Some of them are suitable only for on surfaces like glass or they use legs for locomotion, which results in slow movement. For climbing on concrete surfaces an active vacuum system driven by wheels seems to be the best solution, because of fast continuous motion and a simple mechanical structure. 2. ADHESION CONCEPT The adhesion concept of the climbing robot Cromsci consists of seven single vacuum chambers which are supported by one large low pressure reservoirs at the top of the robot. Each chamber receives its low pressure from the reservoir chamber which is evacuated by three vacuum pumps and controlled by a system of several valves. If one or more working chambers are loosing pressure they can be isolated from the vacuum system to avoid the adjustment to normal pressure. A complete adjustment will result in a complete loss of adhesion. The air-pressure in each chamber and the reservoir is measured by pressure sensors. This information is given to a pressure controller which opens and closes the valves to evacuate the chambers separately depending on the actual leak tightness. The components of the adhesion system are: - One large reservoir chamber (including valve to outside) - Three vacuum pumps evacuating the reservoir chamber - Seven working chambers (each including valve, pressure sensor and sealing) - Electronics for low-pressure control Figure 1: Principe of the adhesion Concept Beside the real adhesion mechanism a thermo dynamical model has been created to simulate the airflow and pressure variation with modelled leakage areas and cracks. It shows that the 2 / 10

3 usage of multiple sucking chambers is as important as a good driving strategy if the cracks are too large for the sealing. More details on the vacuum control system can be found in [5] Sealing Concept An important aspect of the seal concept is the partition between the low pressure inside the vacuum chambers and the normal pressure outside. To guarantee that the climbing-robot keeps the contact to the wall, while moving on its surface, it is necessary to develop a seal which needs to be wear-resistant, leak-proof and easily sliding. To reach this intention a prototype has been developed which enables to apply several different sealing materials and shapes to find the optimum in friction and wear General operating mode of the adhesion concept In Figure 2 the general operating mode is displayed: Figure 2: General operating mode of the adhesion Concept and Forces The two functions of propulsion and sealing are separated. This is necessary, because the sealing cannot stand the full normal force to the wall to hold the whole robot. It only needs enough down force to the wall to keep the seal in contact and isolate the vacuum chambers. Only by dividing the two basic functions, it is possible to create a sliding sealing-system. The normal force, which presses the robot to the vertical surface, has to be absorbed by the propulsion system. The right picture of Figure 2 illustrates the effective forces on the climbing Robot. For fitting on the wall a low pressure in the suction chambers is essential. For a motion of the robot the friction force of the wheels must be higher than the sum of the gravity force of the robot and the friction force of the sealing. Therefore a small pressure force of the sealing is profitable Flexible constitution of the sealing mount Because of the stiff propulsion and the stiff housing, a flexible constitution of the sealing mount is essential, since it is not possible to balance great irregularities with the structure of the sealing surface. On a real wall humps and holes occur frequently in many different variants. In Figure 3 an overcome of a irregularity in the wall is exemplified in three steps. 3 / 10

4 Figure 3: Overcome an error on the surface 2.2 Basic Research In the following chapters the possibilities of testing the sliding surface will be presented. The tests are divided in two main parts. The first part describes the basic tests in which only the friction between several test materials is compared. In the second part the friction charged with low pressure and air-stream in the contact is tested. With the second test, the real circumstances of the adhesion system in the area of the sealing can be simulated Test Rig Pin on Disc To detect the lift/drag ratio and the value of friction between the sample and the surface vis-àvis, a Pin-on-Disc test rig is used which fulfils DIN [6]. With this test rig it is possible to adjust different velocities of the turning disk to simulate the relative movement between seal and wall. The normal force, as result of the vacuum in the adhesion module can be simulated by some weights, which are placed on top of the pin. The sliding friction is detected by using a beam in bending, which is coevally the reference surface for the inductive sensor to detect the amplitude of the beam. To determine the wear across the time of testing, a second inductive sensor is applied on the opposite site of the beam in bending. They are both placed on one rocker, shown in Figure 4. The sliding friction and the wear will be detected while using a dry, wet and steamed contact between pin and disk. In this way, all conditions, which appear on the sealing of the climbing robot, can be simulated. Figure 4: Pin on Disc test stand 4 / 10

5 2.2.2 Friction of different sealing materials In the first studies with the Pin on Disc test rig the friction between several samples of seals are tested. In the following chart, the different parameters are listed: Test No. rel. Velocity Temperature Lubrication Contact Pressure Material Pin Material Disc [m/s] [ C] 10-3 x[n/mm 2 ] 1 0,1-0, ,4 Rubber Glass 2 0,1-0, ,4 Rubber Marble 3 0,1-0,03 20 Dust 9,4 Rubber Concrete 4 0,1-0, ,4 Carpet Glass 5 0,1-0, ,4 Carpet Marble 6 0,1-0,03 20 Dust 9,4 Carpet Concrete Figure 5: Parameters for first test with pure friction The tests no. 1 to 3 are made with rubber on glass, plane marble and plane concrete. They provide a basis to compare well known materials with the samples of fleece (carpet). In all experiments the temperature and the contact pressure were kept constant. It is possible, that the temperature departs in a field of +/- 1 Kelvin but this is irrelevant for the result of the friction coefficient. It was attempted to run the test without any kind of lubrication, but this was not possible with the samples of concrete. The concrete disc wears fast, because the pin in contact grinds the fine sand on the surface. This fine sand works like a kind of lubrication and tempers the dry friction. However all pins, which stay in contact with the concrete disc, excite this special wear, so they are comparable among each other. Figure 6: used Discs Three different materials were used for the pins: rubber, fleece 1 (short fibres), fleece 2 (long fibres). Summary of results: Figure 7: used Pins Test No. Static Friction Dynamic Friction 1 rubber - 1,1 1,7 2 rubber 2,3 2,7-3 rubber 3,7 4,9-4 fleece 1-0,33 0,37 4 fleece 2-0,20 0,25 5 fleece 1-0,17 0,25 5 fleece 2-0,32 0,37 6 fleece 1 0,70 1,00-6 fleece 2 0,60 1,22 - Figure 8: Results of the pure friction tests (graphs in [7]) 5 / 10

6 2.2.3 Tests with air-stream in the sealing contact In the following tests, a new fleece with long and close set fibres was analysed with an additional load of air-stream and small vacuum in the contact. The new fleece is chosen, because it seems to be the most appropriate friction layer to use for the sliding contact. Compared to the fleeces used in the first tests, the compact new probe has a lower friction coefficient but with a similar sealing performance as the other materials. The fleece combines the high concentration of thin strait fibres of the short hair fleece and the sealing potential of the long hair fleece, with its smooth fibres. Figure 9 shows the adjusted pin with the low pressure area in the middle of the fleece probe. The results of the new pin cannot be directly compared to the results of the tests of only friction without streaming in the contact. The variation of the geometry of the pin probe causes different sliding circumstances in the contact which causes variances in friction. To compare the new fleece to the ones used in the first test, all probes need to have the same geometries. Figure 9: New fleece with a new geometry of the pin The streaming tests are made to simulate the circumstances in the low pressure chamber of the climbing robot. Only with a realistic air-stream in the sealing contact and the low pressure in the vacuum chambers, it is possible to simulate the sealing and form an opinion about the quality of the contact partners. The first tests could only show the direction, which sliding material could be practical, but it cannot simulate the sealing concept. The parameters of the streaming test are chosen as a result of the estimated situation, while the climbing robot is in action. Parameters for the streaming tests Test No. RPM Difference in Pressure Temperature Normal Force without Pressure Lubrication Material Pin Material Disc [kpa] [ C] [N] ,84 - New Fleece Glass ,84 - New Fleece Marble ,84 Dust New Fleece Concrete Figure 10: Parameters of the tests with the new fleece 6 / 10

7 Figure 11: Friction force fleece to glass, comparing with and without air-stream The fleece pin in combination with the glass disc causes a considerable stick slip effect, so the maximums of the amplitudes have to be compared. Figure 12: Friction force fleece to plane marble, comparing with and without air-stream The courses of the graphs in Figure 12 are similar to Figure 11, but the amplitude of the test with the air-stream is a little higher and not as constant. The irregularities, which can be seen as a variation of about 0,3 Newton of the amplitude, has to be caused by small irregularities on the disc surface. Marble is a natural material and has a high number of scratches and cracks. Plane and polished it is similar to glass, but the natural defect in the surface still causes a variation of the friction force. Figure 13: Friction force fleece to concrete, comparing with and without air-stream 7 / 10

8 The test with the concrete disc differs from the other tests with the air-stream contact. The amplitude of the friction force is similar to the amplitudes of glass and plane marble, but the frequency of the vibration caused by the irregular surface of the concrete, is not comparable to the other tested discs. The dust, which occurs while the pin slides over the surface, seems to lubricate the contact and thereby reduces the friction force and vibration amplitudes. The course of the graphs is very constant and does not depend on the velocity between pin and disc. Summary and results of the streaming test: Test No. Contact Friction Force with normal Pressure [N] Friction Force with low Pressure [N] Factor: low to normal Pressure Friction Coefficient 1 Fleece-Glass 0,82 1,4 1,70 0,29 2 Fleece-Marble 0,85 1,85 2,18 0,30 3 Fleece-Concrete 0, ,61 0,32 Figure 14: Results of streaming tests 3. DESIGN During the development process the design of the sealing was subject to many changes. The different stages of development are presented below: 3.1. Pressurized Sealing Mount In a first step a clamped elastomeric torus is used as the mount of the friction surface. By varying the pressure, the working height can be adjusted. In Figure 15 the first variation of the pressurized sealing mount is shown. Figure 15: Pressurized sealing mount (without separate sealing surface) To realize a separate sealing surface with an adequate friction coefficient, the following variants were realized: in Figure 16 a) a fold up friction pad is connected to the pressurized mount. With this assembly the height of the sealing can be adjusted by the pressure. This assembly is characterised by a low stiffness against height adjusting. Figure 16: Fold up friction layers a) and b) 8 / 10

9 In the second model a friction tube is connected to a rigid mount, like shown in Figure 16 b). By increasing the pressure in the bigger tube in the middle of the mount, an extra function is realized: In fact of loosing the connection to the wall, the tube can be extended, so that the vacuum can be restored by closing the leak. This assembly is characterized by a very small height adjusting Convoluted Rubber Gaiter The versions discussed above were not suitable for the climbing robot. In the next step a selfadjusting convoluted rubber gaiter is used, whereof three different variants were realized: Plumber s helper Figure 17: Plumber s helper and convoluted rubber with friction surface The variant Plumber s helper is pictured in Figure 17. With a plumber s helper as a mount for the sealing surface, a prototype is built up. The socket causes adhesion, if low pressure exists. In a first test the socket adhered, so a movement was not possible. Based on the experience with the plumber s helper a convoluted rubber gaiter was crafted (Figure 17). In this model the affection to adhere is smaller, due to the lower stiffness in axial direction. The small-scale variant was tested on the Pin on disc test rig and thereby considered as a suitable variant Full-Scale Convoluted Rubber Gaiter With this gain in experience the next variant is a full-scale convoluted rubber gaiter, like shown in Figure 18. Its dimensions are similar to the geometry of the vacuum chambers of the robot to obtain a realistic impression of the forces, which will occur in real action. Figure 18: Full-scale convoluted rubber gaiter with friction surface 9 / 10

10 4. CONCLUSION In this paper a new sealing system was introduced, which seems to be very promising. Several sealing prototypes were presented and different applicable facings have been tested regarding friction and abrasion on several surfaces like concrete, marble or glass. These experiments pointed out, that the sliding characteristic strongly depends on the facing materials and will have a high influence on the movement and adhesion of the sealing system. Furthermore, the influence of low pressure and air-stream could be presented in the second part oft the experiences on the pin on disc test rig. Next steps and future work mainly consist of experiments under real conditions in a real environment. For these tests a next sealing prototype was already built. It consists oft seven parts, each for every chamber, which was developed on the basis of the prototypes, shown in chapter 3. Figure 19: Sealing for real tests in real environment upside down In addition to that more experiments concerning the leak tightness of the sealing facings have to be carried out. These results will be compared to simulation results and will then be evaluated in order to achieve best adhesion attributes. Books / References: [1] Y. D. Zhang, K. L. Fan, B. L. Luk, Y. H. Fung and S. K. Tso, Mechanical design of a climbing cleaning robot, in 8th IEEE Conference on Mechatrinics and Machine Vision in Practice, (Hong Kong, 2001). [2] D. Longo and G. Muscato, Design of a single sliding suction cup robot for inspection of non porous vertical wall, in ISR 2004: 35th International Symposium on Robotics. Proceedings., (Paris, France, 2004). [3] M. Haegele, Assistive robots in everyday's environments Europe's Information Society - ICT Riga 2006(June, 2006). [4] B. L. Luk, D. Cooke, S. Galt, A. A. Collie and S. Chen, Robotics and Au-tonomous Systems 53, pp.42(october 2005). [5] J. Wettach, C. Hillenbrand and K. Berns, Thermodynamical modelling and control of an adhesion system for a climbing robot, in 20th IEEE International Conference on Robotics and Automation (ICRA), (Barcelona, Spain, 2005). [6] Oenorm din 53516: Testing of rubber and elastomers - determination of abrasion (1987). [7] C. Hillenbrand and K. Berns, The force controlled propulsion and adhesion system for a climbing robot, in 9th International Conference on Climbing and Walking Robots (CLAWAR), September / 10