MECHANICS OF PNEUMATIC ARTIFICIAL MUSCLE

Transcription

1 MECHANICS OF PNEUMATIC ARTIFICIAL MUSCLE *Zdeněk Varga, **Michal Moučka *Technical University of Liberec, Department of Applied Cybernetics Studentská 2, 46117, Liberec 1, Czech Republic **Technical University of Liberec, Department of Applied Cybernetics Studentská 2, 46117, Liberec 1, Czech Republic Phone: , Fax: , Artificial muscles are used, for example, for artificial hands of robots or positioning of single mechanisms. Their lower mass and faster response to the change of air pressure are appreciated. The aim of this article is to present results gained in research experiments with an artificial muscle, which will be used for development of a theoretical model. This model can be later used for direct control of the muscle. Within the experiments it was confirmed that the pneumatic artificial muscle is non-linear system and temperature is very important variable which has very big influence on muscle's behaviour. All measurements were performed on a newly built stand in the laboratory on the Department of Applied Cybernetics. Experimental data were compared with theoretical models from literature. Keywords: pneumatic artificial muscle;experiment; model 1 INTRODUCTION Artificial muscles are known since 60-ties of 20th century and since 90-ties they are manufactured in serial production. They are used for both simple and complicated mechanisms, for example to such as artificial hands of robots or positioning of single mechanisms. In these applications two main features of the muscles are used - their lower mass and also faster response to the change of air pressure source. The presented paper deals with results of research experiments with artificial muscles at Department of Applied Cybernetics at Technical University in Liberec. The realized experiments were focused to finding the exact mechanical features of artificial muscle. The obtained parameters will be used for mathematical models of muscles. The mathematical models will be used for design of algorithms for positional and velocity control of the muscles. 2 EXPERIMENT SETTINGS For the purpose of research of mechanical, static and dynamic properties of the pneumatic muscles the measuring stand was compound. Figure 1: The scheme of measuring stand In the figure 1 the scheme of stand is shown. The basic elements of the stand is the pneumatic muscle FESTO MAS N-AA-MC-O-ER-BG, the pneumatic electro-proportional pressure valve FESTO VPPM-6L-L L10H-A4P-S1C1 and fast precise distance laser sensor FESTO SOEL-RTD-Q50-PP-S-7L. The length of the muscle depends on pressure which is in its chamber. The size of the pressure is controlled by means of the proportional valve. 1

2 The actual length of the muscle is measured by mentioned distance laser sensor. The size of the pressure in chamber of the muscle can be measured by the pressure sensor, which is a part of the proportional valve. The compressor Orlík SKS9/200 is the supply of pressured air. The supplied air is dehumidified in the condensation drier Orlík KSO 010A and filtered by the micrometrical filter FESTO LFR 1/4-5M-MIN. In the measuring device there is also inserted the electro-pneumatic stop valve FESTO HEE 1/4-MINI-24. The data acquisition card NI (National Instruments) 6252 PXI, which is used for measuring and controlling of the electro-pneumatic proportional valves is included in industrial computer. The measuring process was realized by means of an application which was developed in the programming environment LabVIEW of version 8.6 (see figure 2). Figure 2: The application for measuring Figure 3: The measuring stand Hanging of the muscle was purposely chosen as vertical (see figure 3) instead of antagonistic (i.e. two parallel muscles) because of the elimination the friction. The friction is undesirable. We want to avoid any non-linearities in the system. Within the experiment, these characteristics: contraction dependence on pressure and loading; hysteresis during loading; measuring of warming up of the muscle during loading were measured by the measuring stand. Mechanical properties of the pneumatic muscle have influence on these characteristics. In the next chapter measured characteristics are described. 3 EXPERIMENT RESULTS 3.1 Measuring contraction and hysteresis of the artificial muscle a) Dependence of muscle contraction on pressure Figure 4: Graph of hysteresis 2

3 The measurement results show that the artificial muscle has a hysteresis and the loading characteristic (figure 4) is non-linear. b) Dependence between pressure and contraction with variable loading The hysteresis on the muscle was measured in a cycle with variable loading. Figure 5 shows that contraction in dependence on pressure grows with loading. It can be simply said the figure 5 was created from the figure 4 by the addition of the third dimension which is loading. a) Pressure growing b) Pressure lowering Figure 5: Contraction dependence on pressure and loading The graphs of loading in the figure 5 is split up into two parts due to better visibility. It might be noticed that the muscle behaviour is non-linear both in pressure growing / lowering and as well in change of the nominal loading. This can be seen in both graphs a), b) upper limit lines are not parallel with bottom limit lines. 3.2 Measuring change of temperature during progress All known models speak only about temperature constant model of the artificial muscle in spite of the fact that the muscle is worming up during progress. However, temperature of the material must not be ignored within the mathematics model development because the temperature of muscle material rises during the loading. The muscle was produced from anisotropic hyperelastic material. This material is distinguished by temperature dependency. a) Measuring stand for temperature measurement b) Variable temperature along muscle's length Figure 6: The measuring stand with the resistance wires 3

4 The change of the temperature has influence on hysteresis of the load characteristics and mechanical properties of the muscle. Thus, to be able to develop mathematical model of the muscle within the next research the temperature of the muscle during progress with defined loading was measured. To do that, the measuring stand had to be modified for temperature measuring of the muscle. Temperature measurement was performed by means of resistance wire. It was placed on top and bottom tail of the muscle. The placement of the wires is visible in the figure 6, which was obtained from thermo-camera. From the thermo-camera pictures there was made a series of graphs of the muscle temperature. One of them is shown in the figure 7. a) Temperature before measuring b) Temperature after 2 minutes of work Figure 7: The temperature measuring on the muscle The temperature measuring showed that temperature along the muscle is no constant. Temperature at the top tail of the muscle is lower than at the bottom tail. All graphs were put together into one three dimensional chart, see figure 8. Obtained knowledge of warming up of the artificial muscle will be used for creating of the mathematical model in future. Figure 8: Three dimensional graph of temperature measuring 4

5 3.3 Volume model of artificial muscle The volume of the artifical muscle depends of air pressure in its chamber. For description of this depecdence a lot of teorethical models exists. Three common models are widely used. First of them considers border thickness of the muscle and the second model does not consider it [1], [2]. Third one by Klute, G. Klute [5] describes muscle behaviour by using of material features. Within this article measured data were compared with the first model mentioned above. The volume V of the muscle in accordance with [1] is approximately equal to V = 1 D2 L, 4 (1) where D is diameter of the muscle, L is length of the muscle. The size of lenght L can be calculated by the equation L = b cos, (2) where b is thread length and θ is angle between braided threads. The textile threads are placed in the rubber body of muscle. They form a mesh where threads are mutually crossed. The process of mesh production as called as a braiding. For diameter D next equation is valid D = b sin. n (3) The coefficient b is length of the thread and n is number of turns of the thread. The angle θ was measured on the small sample of cut muscle. In the next figure 9 the way of the measurement is shown. The average value of the braided angle θ on muscle was From braided angle the length b of thread was calculated for known length of the muscle. The length of thread is constant. It means that it can be used for calculation of the muscle volume V, which depends on air pressure in its chamber. Then theoretical length b of the thread is calculated from known length L of the non-loaded muscle and from average angle according to equation (2). The result is m. Furthermore, the theoretical number of turns n of the thread was determined as from equation (2). Figure 9: Measuring of braided thread angle The theoretical volume V of the muscle in dependency of its length L was calculated for selected level of air pressure. The calculation was provided twice. In the first case, muscle thickness was included, in the second case, thickness was neglected. The volume dependency on pressure was also specified by a experiment. The muscle was sank into the wather vessel. After set of pressure, the length of the muscle and change of the wather-level was measured. The change of the wather-level corresponding with change of the volume of the muscle. Obtained data was used in course of accuracy verification of the Chou Hannford model. Results from volume calculation and volume measurement are shown in the figure 10. Charts in figure 10 shows that measured volume is equal to the theoretical value of the first model (with border thickness) at the beginning of progress, i.e. without presssure. With higher pressure the measured volume gets closer to second model (the one without border thickness). It can be said that the both Chou models cannot be used for modelling of the muscle behaviour without additional corrections during the progress. 5

6 Figure 10: The comparison of the measured and theoretical volume 4 CONCLUSION The experiments with artificial pneumatic muscle performed at the Department of Applied Cybernetics at Technical University in Liberec showed that: the muscle has non-linear behaviour both in pressure growing / lowering and as well in change of the nominal loading, temperature progression along the muscle is non-linear, which influences mechanical features of the muscle (rubber), theoretical models (not)considering border thickness cannot be used for control of the muscle without additional corrections during the progress. Obtained knowledge will be used for development of a new model of the muscle in the following research. REFERENCES [1] CHOU, P., HANNAFORD, B.: Measurement and Modeling of McKibben Pneumatic Artificial Muscle. IEEE TRANSACTIONS on Robotics and Automation. 1996, vol. 12, no 1, p. [2] TSAGARAKIS, N., CALDWELL, G. D.: Improved Modelling and Assessment of pneumatic Muscle Actuators. In Proceedings of International Conference on Robotics & Automation. San Francisco. Dept. Of Electronic Engineering University of Salford Manchester M5 4WT, pp [3] DAVIS, S., CALDWELL, G. D.: Braid Effects on Contractile Range and Friction Modeling in Pneumatic Muscle Actuators. The International Journal of Robotics Research. 2006, vol. 25, no 4, p. [4] SCHULTE, H. F.: The Application of External Power in Prosthetics and Orthotics. National Academy of Sciences-National Research Council. Washington, D.C p. [5] KLUTE, G. K.: Actuators for Biorobotic Systems. Washington, D.C p. University of Washington ACKNOWLEDGEMENTS The work presented in this paper was supported by the research project of Ministry of Education of the Czech Republic under contract code MSM