Improved Design of a Three-degree of Freedom Hip Exoskeleton Based on Biomimetic Parallel Structure

Transcription

1 Iproved Design of a Three-degree of Freedo Hip Exoskeleton Based on Bioietic Parallel Structure by Min Pan A Thesis Subitted in Partial Fulfillent of the requireents for the Degree of Master of Applied Science in The Faculty of Engineering and Applied Science Mechanical Engineering Progra University of Ontario Institute of Technology July, 2011 Min Pan, 2011

2 Abstract The external skeletons, Exoskeletons, are not a new research area in this highly developed world. They are widely used in helping the wearer to enhance huan strength, endurance, and speed while walking with the. Most exoskeletons are designed for the whole body and are powered due to their applications and high perforance needs. This thesis introduces a novel design of a three-degree of freedo parallel robotic structured hip exoskeleton, which is quite different fro these existing exoskeletons. An exoskeleton unit for walking typically is designed as a serial echanis which is used for the entire leg or entire body. This thesis presents a design as a partial anipulator which is only for the hip. This has better advantages when it coes to arketing the product, these include: light weight, easy to wear, and low cost. Furtherore, ost exoskeletons are designed for lower body are serial anipulators, which have large workspace because of their own volue and occupied space. This design introduced in this thesis is a parallel echanis, which is ore stable, stronger and ore accurate. These advantages benefit the wearers who choose this product. This thesis focused on the analysis of the structure of this design, and verifies if the design has a reasonable and reliable structure. Therefore, a series of analysis has been done to support it. The obility analysis and inverse kineatic solution are derived, and the Jacobian atrix was derived analytically. Perforance of the CAD odel has been checked by the finite eleent analysis in Ansys, which is based on applied force and oent. The coparison of the results fro tests has been illustrated clearly for stability ii

3 and practicability of this design. At the end of this thesis, an optiization of the hip exoskeleton is provided, which offers better structure of this design. iii

4 Acknowledgents Working on this thesis took lots of tie and effort, but the process has been both interesting, enjoyable and was a great learning experience. The reason for that is I have been working alongside any knowledgeable and intelligent colleagues. I would like to thank the for their great dedication and support for e during this period. First and foreost, I would like to express y deepest gratitude to y supervisor, Dr. Dan Zhang, for his continuous support, encourageent and guidance throughout the research and y graduate study. I would also like to thank hi for allowing e and giving e the opportunity to use and work in the excellent Robotics and Autoation Laboratory. Dr. Zhang has broad knowledge, percipience and otivation on all his research projects, which encouraged e throughout y study. I would like to thank hi for his valuable supervision, full support and review of the anuscript. This work would and could never be what it is like right now without his contribution. In particular, I also offer thanks to Mr. Zhongzhe Chi, who has been played an iportant role in the way I approached y Master degree. His invaluable advises and guidance always get e out fro bottlenecks. I would not have been able to solve the probles I have encountered without his assistance with his insights and experience. Also y sincerest thanks to all the ebers of the Robotics and Autoation Laboratory at UOIT for their intelligent advises and kind encourageent. They are such a group of lovely people, who always share ideas and inforation with each other. They are always ready to lend a help hand to others regardless in acadeia or life. It has been a pleasure iv

5 and a privilege to study in such a war environent though the weather in winter is really cold. Last but not least, I would like to give y deepest thanks to y Mu, Dad and brother for always being there for e and keeping y ego in check. I would not be able to continue y graduate study without their financial and support. And finally, y thanks to the ost special an in y life for his patience and understanding as well as his effort in pushing e. v

6 Table of Contents Abstract... ii Acknowledgents... iv Table of Figures... viiiiii Table of Tables...x Chapter 1 Introduction Literature Review Subject of the Study Motivation The organization of the Thesis...9 Chapter 2 Bioechanical Gait Analysis Regular Huan Gait Cycle Kineatics and Kinetics of the Gait Muscle Action in Gait Cycle...17 Chapter 3 General Design of the Hip Exoskeleton Unit The Purpose of the Hip Exoskeleton Unit Requireents on the Structure Requireents on Applicability Requireents on Econoic Consideration Hip Exoskeleton Geoetric Structure Position Analysis CAD Design Kineatics Analysis Mobility Analysis Inverse Kineatics Analysis Jacobian Matrix Stiffness Analysis Sensing and Controlling of the Syste Sensor Selection and Placeent Control Motor Selection and Placeent...46 vi

7 3.5 The Hip Exoskeleton Interface to the Wearers...48 Chapter 4 Syste Optiization FEA Analysis Force Moent Workspace Evaluation Stiffness Optiization Achievable Iproveent of Exoskeleton Developent on Geoetry Developent on Cofort Level...69 Chapter 5 Results and Discussion...74 Chapter 6 Conclusions...77 Chapter 7 Future Work Prototype Materials Sensor and Motor Further Test and Optiization...81 Reference...83 Appendix...86 Appendix I: FEA Analysis of the Iproved Structure...86 Appendix I.I: Force Application...86 Appendix I.II: Moent Application...88 Appendix II: FEA Report of the Iproved Structure...91 vii

8 Table of Figures Figure 1.1: The Basic Structure of a Serial Robotic Manipulator... 2 Figure 1.2: The Basic Structure of a Parallel Robotic Manipulator... 3 Figure 1.3: CNC Mechanis Tool... 4 Figure 1.4: Cyberdyne Hal Figure 1.5: Berkeley Lower Extreity Exoskeleton... 5 Figure 1.6: Spring Walker... 6 Figure 1.7: Murdered Professor s Walking Aid... 7 Figure 2.1: Principal Planes of the Huan Body in a Standard Anatoic Position Figure 2.2: Eight Phases in a Gait Cycle Figure 2.3: A Noral Gait Cycle Figure 2.4: Sagittal Plane Hip Motion Figure 2.5: Sagittal Plan Thigh Motion Figure 2.6: Moent and Power Profiles of Hip Motion Figure 2.7: Muscle Action during Gait Figure 3.1: CAD Model of the Designed Hip Exoskeleton Figure 3.2: Prisatic Joint Figure 3.3: Universal Joints Figure 3.4: Siplified Scheatic Diagra of the Hip Exoskeleton Unit Figure 3.5: A Siplified Lib of the Manipulator Structure Figure 3.6: Stiffness Results in Z direction of the Hip Exoskeleton Figure 3.7: Stiffness Results in Y direction of the Hip Exoskeleton Figure 3.8: Stiffness Results in X direction of the Hip Exoskeleton Figure 3.9: Theory of Sensing and Controlling Figure 3.10: Surface EMG Signal Processing and Data Processing Circuit Figure 3.11: Placeent of SEMG Sensor Electrodes Figure 3.12: The Location of Servo Motors Figure 3.13: Servo Motor Working Principle Figure 3.14: The Structure of Liit Switch Figure 3.15: CAD Model with two Leg Exoskeletons Figure 4.1: Finite Eleent Model of the Hip Exoskeleton viii

9 Figure 4.2: Total Deforation based on Applied Force Figure 4.3: Deforation in X, Y, Z-axis based on Applied Force Figure 4.4: Equivalent Stress based on Applied Force Figure 4.5: Equivalent Elastic Strain based on Applied Force Figure 4.6: Total Deforation based on Applied Moent Figure 4.7: Deforation in X, Y, Z-axis based on Applied Moent Figure 4.8: Equivalent Stress based on Applied Moent Figure 4.9: Equivalent Elastic Strain based on Applied Moent Figure 4.10: Workspace Volue of the Hip Exoskeleton Figure 4.11: Workspace Volue Optiization Results Figure 4.12: Upgrading CAD Model of the Hip Exoskeleton Figure 4.13: Lower Leg Coparison between Two CAD Modelsl Figure 4.14: Joints Coparison Between Two CAD Models Figure 4.15: Total Deforation of Updated Model based on Applied Force Figure 4.16: Lower Leg Belt with an Inner Gibal at Static Position...70 Figure 4.17: Lower Leg Belt with an Inner Gibal at Moving Position Figure 4.18: Lower Leg Belt with a Spherical Gibal at Static Position...71 Figure 4.19: Lower Leg Belt with a Spherical Gibal at Moving Position..72 Figure 5.1: Move Forward.74 Figure 5.2: Move Backward Figure AIII.3: Move Right Figure AIII.3: Move Right Figure AI.1: Finite Eleent Model of the Updating Hip Exoskeleton Figure AI.2: Total Deforation based on Applied Force Figure AI.3: Deforation in X, Y, Z-axis based on Applied Force Figure AI.4: Equivalent Stress based on Applied Force Figure AI.5: Equivalent Elastic Strain based on Applied Force Figure AI.6: Total Deforation based on Applied Moent Figure AI.7: Deforation in X, Y, Z-axis based on Applied Moent Figure AI.8: Equivalent Stress based on Applied Moent Figure AI.9: Equivalent Elastic Strain based on Applied Moent ix

10 Table of Tables Table 3.1: Suary of Huan Data for Design Table 3.2: Hip Noral Range of Motion Table 3.3: The Designed Hip ROM Table 3.4: Critical Diensions of the Manipulator Table 4.1: Optial Design Paraeters for Maxiizing Workspace Volue x

11 Chapter 1 Introduction Currently, robots are not far away fro people s life, such as a surgery ar, service robots, entertain robots, etc. According to the Webster s New World College Dictionary, a robot is any anthropoorphic echanical being built to do routine anual work for huan beings. However, the Robotics Institute of Aerica gives a ore general definition a robot is a re-prograable ulti-functional anipulator designed to ove aterial, pars, tools, or specialized devices, through variable prograed otions for the perforance of a variety of task [1] To siplify it, a robot is a echanis that is able to perfor tasks on its own or with control [1]. In the past few years, the deand for high perforance robots for daily huan activities increased rapidly due to the advanceent of robotics technology. Robots are classified into different categories based on various criteria. There are soe coon criteria used to classify the robots: the size of the robots icro robots and acro robots; the drive technology electric, hydraulic, and pneuatic; the workspace geoetry; the degrees of freedo; the otion characteristics; and the ost coprehensive one, kineatic structure serial robots, parallel robots, and hybrid robots. A serial structured robot is also called an open-loop anipulator which is consisted by a nuber of links connected with either a prisatic or a revolute joint in an 1

12 open kineatic chain [2] as shown in Figure 1.1 [3]. The ost significant advantage of a serial robot is the large workspace, so it is able to take up a sizable work area easily. Serial anipulator is one of the ost coon industrial robots due to its large volue occupation. That is because it can be used to carry relatively heavy load or do heavy duty tasks. Therefore, serial robots are the ost basic ones selected for heavy load designs or projects. Figure 1.1: The Basic Structure of a Serial Robotic Manipulator [3] Different fro a serial robot, a parallel robot has two platfors which are connected by at least two kineatic chains, and the otion is achieved by the endeffector platfor [2] as shown in Figure 1.2 [5]. Parallel robots are usually selected to satisfy the need for high accuracy tasks because they have higher operational accuracy, broader range of load capacity, better task flexibility, better reliability and shorter cycle tie than that of serial robots [4]. On the other hand, parallel robots have uch ore coplex operation than that of serial ones, and they have liitation on workspace. 2

13 Overall, the selection of the robotic structure is based on the requireents of the applications. Figure 1.2: The Basic Structure of a Parallel Robotic Manipulator [5] In the past few years, ore and ore parallel echaniss are chosen because of their high dynaic perforance and high stiffness. As a result, the developent of the parallel kineatic echaniss is attracting ore and ore researchers and industry copanies. At the sae tie, ore parallel robot structures are utilized in real applications, such as illing achines, and surface grinders. The 5-axis reconfigurable parallel robotic achine in the Robotics and Autoation Laboratory of UOIT is a typical parallel structure based achine tool (Figure 1.3). 3

14 Figure 1.3: CNC Machine Tool Because of the attributes of high stiffness, better accuracy, parallel kineatic echaniss are widely used in bioechanics, such as exoskeleton. There are three principal areas that bioechanics are generally divided into: perforance, injury, and rehabilitation [6]. However, this criterion is not strict soe bioechanics have the future that fits all three areas. Based on verity of requireents, different kineatic structures are picked to build different kinds of exoskeletons. 1.1 Literature Review The idea of using exoskeletons to help people with the daily life dates back to 1890 when an apparatus was used for facilitating working, running, and juping [7]. So far, there are a verity of exoskeletons have been developed for different purposes -- to support or protect the inner skeleton, carry heavy load, and help ankind in the daily life. 4

15 Soe designs have been fully developed and started to launch on the arket, such as Cyberdyne Hal-5 (Figure 1.4) [8], Berkeley Bleek Exoskeleton (Figure 1.5) [9]. Soe of the are still in the testing or research stage, such as Spring Walker (Figure 1.6) [7], and Murdered Professor Walking Aid (Figure 1.7) [11], etc. Due to the large deand, the exoskeleton is going to play a uch ore iportant role later. Figure 1.4: Cyberdyne Hal-5 [8] Figure 1.5: Berkeley Lower Extreity Exoskeleton [9] 5

16 Cyberdyne Hal-5 is a cyborg-type robot that is designed for expanding and iproving the physical capability of the person who wears it [8]. This robot can be separated into two parts: upper body parts and lower body parts, so people can wear half of it for certain activities without carrying extra load. Cyberdyne Hal-5 can perfor ost of the daily activities, such as standing, walking, clibing up and down stairs, holding and lifting, so it can be applied in various fields. Also, it can be used for both indoor and outdoor activities due to its hybrid control syste [8]. So far, it is the ost successful exoskeleton for huan routine tasks. Berkeley lower extreity exoskeleton is ainly designed for the people who have to carry significant loads on their backs with inial effort over any type of terrain along unstructured outdoor paths. This robot has seven degree of freedo per leg, and four of the are powered by liner hydraulic actuator that can provide a large power [9] [10]. The principle behind this exoskeleton is that the linear hydraulic actuators add power to the hip, knee, and ankle joints, and then an internal cobustion engine provides the necessary electric and echanical power for the exoskeleton [7]. As a result, the wearer is able to carry a significantly heavy load. Figure 1.6: Spring Walker [7] 6

17 Figure 1.7: Murdered Professor's Walking Aid [11] The Spring Walker was defined as a spring assisted lower body exoskeleton, which was developed in 1991 [7]. It has the capabilities of running and leaping of all anials, and it allows the wearer to run up to 35ph and leap up to 5 feet into the air [11].Therefore, it can be concluded as a coplex, huan powered, kineatic exoskeleton, which is consisted of a kineatic linkage whose joints incorporated springs [7]. As it can be seen, the huan feet do not touch the ground because the legs of the Spring Walker are in series to the huan. The Murdered Professor s Walking Aid was designed in order to recover fro a disease called Sarcophenia, which causes the lost of skeletal uscle [11]. The power is added on to the lower leg and feet, and assists the leg oving forward and backward. This exoskeleton should be classified into to the rehabilitation bioechanics category. After it is been copleted, it should be able to help a lot of patients. 7

18 1.2 Subject of the Study As entioned above, robots are able to work by theselves after engineers or scientists prograed the, but the fact is that the robots need to be supervised, edited and iproved to satisfy the growth requireents fro tie to tie. It is the sae as exoskeleton. As tie goes by, further and ore specific requireents coe into sight. If taking exoskeleton as an exaple, it is easy to explain. There are soe developed echaniss which can help people ove around, such as a scooter. However, it is not that convenient when people are in a sall space, such as in a roo, so a wearable echanis is required. As a result, Cyberdyne Hal-5 could be a solution. Nevertheless, people want to have soething especial for their hip if they only have hip probles rather than have their entire body or entire leg covered by echanis. In order to eet the requireent, a partial echanis is a better solution. However, it has to be able to have the sae ability as the wearer request, for instance, assisting to do the routine oveents walking. The hip exoskeleton unit introduced in this thesis is designed and analyzed for the people who need special assistant with the hip oveents. Coparing with a whole body echanis, a partial anipulator unit is ore convenient and uch lighter. As a result, this product should take the people who have had hip surgeries, the elderly who cannot ove their legs properly as a target arket. Alost all existing exoskeleton assistant echaniss are designed as a serial structure or a hybrid one, but the above entioned target arket have request of safety rather than speed. Therefore, a parallel structure anipulator is ore suitable due to the advantage of accuracy. 8

19 1.3 Motivation In suary, the thesis presents an iproved architecture to a fully actuated threedegree of freedo parallel structure hip exoskeleton unit. Specifically, the parallel structure of this bioechanics was analyzed, and it has proved that the design of the anipulator is ready to go to the next step - prototype. Also, sensors and otors have been introduced, and this provides a guideline to the future research. More or less, the partial anipulator idea will lead people to consider developing lighter and easier robots to satisfy ore practical requireents. Therefore, these analyses will guide the future research effort. 1.4 The organization of the Thesis This thesis postulates the following questions and attepts to prove the solutions by several ethods. Hypothesis I: It is reasonable to design a parallel structure three-degree of freedo hip exoskeleton unit that helps wearers to ove their legs properly; Hypothesis II: The designed structure is reasonable, reliable, convenient, and state-of-the-art. follows: This thesis consists of seven chapters and the road ap of this thesis process as Chapter 2 introduces the background of soe ters associated with huan walking and huan gait cycle data at the hip. Also, it gives a brief idea of the uscle otion which affects the sensors detection. 9

20 Chapter 3 presents the CAD odel of the hip exoskeleton unit. Kineatic analysis consisted by obility analysis, inverse kineatics analysis, Jacobian atrix analysis and stiffness analysis is fully discussed, which give the detailed answer to the thesis questions. At the end of this chapter, control and actuation that includes sensors and otors are introduced in general. This chapter is the ajor part of the solution of the thesis postulate. Chapter 4 focuses on the optiization of the designed syste. First, a FEA analysis with a coputer progra called ANSYS is brought in to assist proving the structure of the odel. Then, the optiization and potential iproveent of the unit are given. This chapter is also one of the ajor contents to the thesis postulate. Chapter 5 deals with the results and discussion based on the data and coparison obtained in the previous chapters. Chapter 6 brings together the ost iportant inforation discussed and lead to the final conclusion of this thesis. Chapter 7 provides suggestions and a guideline for the next steps and future work that could iprove the design to a higher level. 10

21 Chapter 2 Bioechanical Gait Analysis This chapter presents the background of the bioechanics particular huan walking bioechanics and introduced the uscle otion during a regular gait cycle. The energetic of walking during a walking cycle is described. This inforation provides the requireents for the hip exoskeleton unit. Fro this detail inforation, the specifications of the actuation and control of the proposed exoskeleton are introduced. 2.1 Regular Huan Gait Cycle In order to better understand the relation of huan body s joints, a Cartesian coordinate syste can be used to display relation of a huan body (Figure 2.1), and the three defined planes are useful in bioechanics analysis. As described below: the X-Y plane is called the transverse plane (or horizontal plane) that divides the body into upper and lower parts; the X-Z plane is called sagittal plane (or edian plane) that divides the body left to right; the Y-Z plane is called coronal plane (or frontal plane) that divides the body front to rear [6]. By using this three defined planes, the gait cycle can be analyzed, since walking of huan being is a three diensional otion. In the following discussion, the sagittal plane is the only one taking into the analysis because there is has a uch larger otion coparing with the other two planes [12]. 11

22 Figure 2.1: Principal Planes of the Huan Body in a Standard Anatoic Position [6] According to Dr. Jacquelin Perry [2], gait analysis is consisting of Noral and Pathological Function. There are eight walking phases included in a gait cycle, which can be grouped into two periods: stance and swing [12]. Figure 2.2 shows the outline of eight walking phases in a gait cycle, which also provides a detailed siulation of the individual joint otion in the lib diagra. Figure 2.2: Eight Phases in a Gait Cycle [12] 12

23 Figure 2.3 is ade based on the analysis entioned above and the research has been done for the gait cycle, which shows the relationship between the walking periods and phases, which is also considered as the divisions of the gait cycle. If a full noral gait cycle is set as 100%, each of the phases has a certain part of it, but not equal to each other as shown in the following figure. There are two tasks that should be in the stance period: one is called weight acceptance, and the other is called single lib support; then the swing lib advanceent is copleted in the swing period to coplete the entire gait cycle [12]. In a siple explanation, first, the body is giving an order to ove forward. Then, the body is decelerated and stabilized by the uscles at the leg which includes hip, knee and ankle. At the end of the stance period, a pre-swing step is achieved, that is where the ankle is powered by the weight transferred fro the hip and knee in order to ove the body forward. One thing needs to be entioned here that the pre-swing phase is considered as the last step of the stance period as well as the first step of the swing period. As Figure 2.3 shown, each step has a different tie range. Figure 2.3: A Noral Gait Cycle 13

24 The principle behind the gait cycle could help to extract the requireents of the design that will be discussed in the next chapter. However, due to the purpose of this thesis, only the hip kineatics and kinetics are going to be fully studied in this chapter. 2.2 Kineatics and Kinetics of the Gait As defined, bioechanics are the applications of echanical principles to biological or living systes, which include huans, anials, organs, and cells [6]. Most coonly, bioechanics are considered as the echanics which are applied on huan body. In order to develop the ost adaptable and suitable echanis, the kineatics and kinetics of the otion need to be copletely understood. As entioned above, the ajor otion of the hip during a gait cycle on the sagittal plan. At the sae tie, the hip otion follows an approxiate sinusoidal pattern as shown in Figures 2.4 [6] [12] [13] (this figure is built based on the data obtained fro the reference, the sae as Figures ). This figure also gives the range of the hip otion, which is in the range of -20 to 40 degree. Here the positive otion eans flexion while the negative otion eans extension. Fro the Figures 2.3 and 2.4, the peak hip otion is in the Loading Response Phase and the Mid Swing Phase, which is reasonable because the two legs have the largest distance between the at these two positions. The low point is between the Terinal Stance and the Pre-swing phases, and that was the sae as expected. 14

25 Hip Motion (in Degree) Gait Cycle (in %) Figure 2.4: Sagittal Plane Hip Motion Because the hip exoskeleton end-effector is on the thigh, and the thigh otion is taken into consideration when the anipulator is designed. The thigh otion is siilar to the hip otion, which also follows an approxiate sinusoidal pattern. Coparing the Figures 2.4 and 2.5 [6] [12] [13], the ajor difference between these two otions is the otion range. The thigh otion range is between -25 to 30 degree. The reason for that is the position difference [12]. In other words, the hip exoskeleton should be fine if it satisfies the hip otion requests. 30 Thigh Motion (in Degree) Gait Cycle (in %) Figure 2.5: Sagittal Plan Thigh Motion 15

26 The hip oent and power profiles of the gait cycle can be shown in Figure 2.6 [6] [12] [13]. The hip oent and power can be considered as the cobination of the work fro all different uscles. In this case, there are three eleents counted: extensor, flexor, and thigh. The extensor oent and power are still set as positive, and the flexor oent and power are set as negative. The range of the oent is between -130N. and 130 N., and the range of the power needed for the hip otion is in a range of -100N. to 150 W. For the sae reason, if the oent and the power profile can supply the hip otion, it is big enough to supply the thigh otion. The peak negative power needed is in the phase of Terinal Stance, and the peak positive power needed is in the Pre-swing phase that is called pull-off [13]. Hip Moent (in N.) Gait Cycle (in %) Hip Power (in W) Gait Cycle (in %) Figure 2.6: Moent and Power Profiles of Hip Motion 16

27 The hip exoskeleton unit should be able to accoodate the hip s basic otion of which is flexion/extension, abduction/adduction, and edial/lateral rotation (3-DOF otion). Also, it should be able to supply the power needed for the hip and thigh otion as the profiles show. As expected, the study of the kineatics and kinetics of hip helps to generate the general ideas of the design of the control and actuation parts of the hip exoskeleton. More specific requireents of the kineatics and kinetics of hip will be discussed in the following chapters. 2.3 Muscle Action in Gait Cycle In a gait cycle, the action of the uscles which controls the ajor action of the hip is studied in order to apply the principles. The ajor activities of theses uscles that are considered as flexor/extensor and abductor/abductor are shown in Figure 2.7 [12]. (a) (b) (c) (d) (e) Figure 2.7: Muscle Action during Gait [12] 17

28 In this figure, only the thigh uscles are naed because they are the ones that act on the exoskeleton. At each step, only few of the actually work. In Figure 2.7, (a): Sartorius and Rectus feoris; (b): Biceps feoris; (c) Rectus feoris and adductor longus; (d): Rectus feoris, Biceps feoris, and Sartorius; (e): Biceps feoris and Adductor fagnus [10]. During a noral gait cycle, there are five priary uscles act for flexion/extension and abduction/adduction which are called Recues feoris, Biceps feoris, Sartorius, Adductor longus, and Adductor agnus [14]. As discussed above, there are two periods in a gait cycle: stance and swing. In the stance period, these uscles are extensors and adductors. On the other hand, these uscles becoe flexors during the swing period. As a consequence, a fact is that a walking action can be actuated possibly if the uscle action can be detected and passed to an actuator. In general, the kineatics and kinetics of the hip offers the desire on the otor part of the designed hip exoskeleton, the principle of the uscle activity in gait provides the requireents on the sensor due to the type of sensor is selected for the echanis. 18

29 Chapter 3 General Design of the Hip Exoskeleton Unit 3.1 The Purpose of the Hip Exoskeleton Unit The priary purpose for developing this anipulator is to provide the wearers the assistant they need during noral gait otion. Therefore, certain conditions have to be et at the sae tie in designing and building this unit. The basic requireents on the robotic anipulator -- hip exoskeleton -- are high operational accuracy, adaptability, dexterity, and reliability, which can be grouped in the following three categories Requireents on the Structure Based on the current study, ost of the existing exoskeleton robots are serial structure due to its large workspace and easy to control, and ost of these anipulators do not require a high accurate oveent. For exaple, there is no difference for the people who wear BLEEX when the exoskeleton helps the ove a little bit further or higher. Another significant character of these existing exoskeletons is the large occupation. For this design, the precision of the anipulator is one of the critical criteria. The reason is that it ay not be a ajor issue to have a low accurate oveent exoskeleton 19

30 for a healthy person, but it proves to be a big challenge for the wearers who are injured or need to rehabilitate fro a hip surgery. These people do not need a fast oveent or carrying heavy loads but a ore precise support. In order to keep the safe and cofortable, a parallel structure based design for hip exoskeleton is necessary. To accoodate a noral hip otion during walking, the hip exoskeleton has to be able to perfor otion of flexion/extension, abduction/adduction, and edial/lateral rotation (3-DOF otion). A PUU, (prisatic joint - universal joint - universal joint), structure of the links is proposed to fulfill the desired work. A reasonable and reliable structure is the ost iportant part of this anipulator. To verify the structure, a series of analysis has been copleted after the hip exoskeleton odel is built in SolidWorks Requireents on Applicability Different wearers need different sizes of hip exoskeletons, so that the anipulators can provide a good fit and suit the best. However, it is tedious, expensive and iplausible to have each and every hip exoskeleton custo ade. To coe up with general design specifications that would be suitable for ost people is a bug challenge in the designing process. This is discussed later in greater detail. Also, the sizes of the wearers are very diversified different heights and different weights. A regular hip exoskeleton has to be strong enough for ost of people. Hence, there is a specific requireent on the rigidity of the syste, and it is one of the ost iportant criteria to verify the structure of this design. 20

31 The anipulator is used to assist a regular daily activity walking, so this design does not need to have any feature to assist other activities, such as running, sitting, and juping. Nevertheless, the wearers have to sit soetie while they are wearing the hip exoskeleton. In that situation, the exoskeleton ust be easy to take off or put on Requireents on Econoic Consideration There is wear or tear during the usage, and soe parts ay break before other parts do. In ost cases, it is not worth buying a new suit and should just replace the broken part if the iportant parts of the hip exoskeleton are still in good conditions, so there are requireents on repair and replaceent for certain parts of the anipulator. In short, the designed hip exoskeleton has to be designed to satisfy all the following requireents: 1. A parallel robotic structure is necessary; 2. The assistance robot oves forward, backward, and sideway; 3. High accuracy safety issue; 4. Enough stiffness strong enough; 5. Light weight easier to carry; 6. Adjustable fit ost people; 7. Easy to wear/take off - convenient; 8. Easy to asseble cofort level; 9. Modular design for repair convenience econoic concern. 21

32 3.2 Hip Exoskeleton Geoetric Structure To build the CAD odel, a typical wearer needs to be selected because the huan data is going to be refereed. In view of that a fundaental structure of the hip exoskeleton should be able to sustain ost wearers (reliable for big size people), the selected wear is a thirty years old ale who provides the following body data for the design without taking tolerance into consideration. This data are reasonably close to that of the MIT test participant who was of the sae age [7]. Table 3.1: Suary of Huan Data for Design [7] Eleents Unit Diensions Age year 30 Sex ale - Height c 178 Weight Kg 85 Waist size c 130 Top part of the thigh size c 58 Low part of the thigh size c 50 Leg length c 100 Walking speed /s 0.65± Position Analysis The oveent of the hip for a regular walking otion is considered as a threedegree of freedo (3DOF) otion in three otion planes (x-y plane, y-z plane, and x-z plane). These three planes can be called as transverse plan, coronal plane, and sagittal plane. 22

33 The hip range of otion (ROM) of a noral gait cycle in healthy individuals reported in the literature provides the following data according to the obtained data fro the relevant aterials [6] [14]. Reference/ Unit Roass and Andersson Table 3.2: Hip Noral Range of Motion [6] [14] Flexion Extension Abduction Adduction Medial Rotation Lateral Rotation c c degree degree degree degree 12 ± ± ± ± ± ± 6.8 Roach and 12.1 ± Miles ± 5 42 ± ± 8 32 ± 9 Departents of the Ary Gerhardt and Rippstein Average The above data are referred fro healthy individual; the people who need to wear the anipulator cannot have a larger ROM than it. If the anipulator can achieve the above ROM, it is ore than good enough for a regular wearer. Also, the exoskeleton ROM is different fro the inner-skeleton. Thus, the designed exoskeleton has to eet the following possible position ranges in Table 3.3 in order to help people to achieve the possible average daily activities. 23

34 Table 3.3: The Designed Hip ROM [6] [12] [14] Maxiu Designed Moveent Plane Motion Displaceent Displaceent Forward 60 degree 70 degree Sagittal Plane Backward 30 degree 35 degree Coronal Plane Up/Down 10 c 15 c Outside 45 degree 50 degree Transverse Plane Inside 20 degree 25 degree After a odel is built, a series of coprehensive analysis will be tested to quantify and validate this designed structure, such as obility analysis and inverse kineatic analysis. Furtherore, this design has to take the considerations on otor control, sensors and actuator, and anufacturing aterial CAD Design The priary CAD odel of the 3-DOF P-U-U parallel structure hip exoskeleton has been built with SolidWorks (see Figure 3.1) based on the previous research and analysis has been done. 24

35 Figure 3.1: CAD Model of the Designed Hip Exoskeleton: (1) Motor (2)Waist belt (3) Upper leg (4) Leg joints (5)Upper leg joint (6)Sensor connecting point (7) Upper leg belt (8) Lower leg (9)screw (10) Lower leg joint (11) Lower leg belt This odel is built by taking a regular size person as a diension reference in Table 1. As the figure shows, the top part is the waist belt that will be fixed on patient s waist, and the size of the belt can be adjusted based on the size of waist of the operator. As a result, the full aount weight of the structure will be carried on the patient s waist. This could be a proble for the patients who ay not be able to carry such a heavy structure. The second belt and the top thigh belt are tied on the top part of thigh, which can be adjusted the size as the waist belt. The purpose of this belt is to fix the path of the botto poles. The botto thigh belt is tied on the botto of the thigh, and it senses the 25

36 otion of the leg and actuates the otion to the poles in order to generate the oveent of the anipulator. As a result, three otors and five sensors are needed though there are six poles due to the power needed. Also, there is no need for cable to pass the signals due to the position of sensors. The entire syste can be considered as a non-cable-driven parallel echanis. Based on diensions of an average size person, the ost iportant diensions of the structure (Table 3) are odeled by calculating the displaceent and otion output of the work platfor or the otion of the legs. These diensions are fro the CAD odel that introduced above, and will be used for further analysis. Table 3.4: Critical Diensions of the Manipulator Part Naes Diensions Units Inner waist 1300 Outer waist 1420 Thickness of waist 5 Length of lower leg (l) 200 Thickness of the upper leg 20x30 Length of upper leg 180 Diaeter of the lower leg 20 Inner of the upper thigh length (R) Inner of the lower thigh length (r) Diaeter of the belt joint 20 Height of the belt joint 40 Angle between the upper leg and the leg belt (θ) 70 degree Angle between each leg 120 degree The weight of the hip exoskeleton is not deterined due to the variety of the aterials used for each part. 26

37 3.3 Kineatics Analysis The structure of the entire syste is verified by copleting the following kineatics analysis Mobility Analysis The degrees of freedo (DOF) of a echanis are the nuber of independent paraeters or inputs needed to specify the configuration of the echanis copletely. The DOF of a echanis or obility deterination can be exained by using the Chebychev-Grübler-Kutzbach s Forula [15]: g M dn ( g 1) f (1) i 1 i Where: M : Degrees of freedo of a echanis, d : DOF of the space in which a echanis is intended to function ( 3 for planar otion, and 6 for spatial otion), n : Nuber of links in a echanis, including the fixed link, g : Nuber of joints in a echanis, assuing that all joints are binary, f i : Degrees of relative otion peritted by joints. This parallel hip exoskeleton can be siplified as a 3 P-U-U (prisatic, universal, and universal) joints anipulator. The three otion poles actuated by the prisatic joints which are connected to the otor (Figure 3.2), and pass the signal to the end effectorplatfor by the two universal joints (Figure 3.3). A prisatic joint has one degree of 27

38 freedo and iposes five constraints between the paired eleents, and a universal joint has 2 degree of freedo [2]. As a result, each lib of the anipulator has: f =1+2+2=5 DOFs (2) i Figure 3.2: Prisatic Joint Figure 3.3: Universal Joints The siplified hip exoskeleton can be shown as a scheatic diagra in Figure 3.4. By analyzing this scheatic diagra, we can get the DOF of the hip exoskeleton is three by applying the Chebychev-Grübler-Kutzbach s forula, With d =6; n = 8; g = 9; f i = 5, (3) 28

39 Figure 3.4: Siplified Scheatic Diagra of the Hip Exoskeleton Unit Fro Figure 3.4, we can see that this anipulator has two platfors, the base platfor, B 1 B 2 B 3, and the end-effector oving platfor, A 1 A 2 A 3. The coordinate axis of the inertial frae is labelled as O B -X B Y B Z B at the center of the base platfor, and Y B - axis is aligned with the point B 2. The oving frae denoted the coordinate axis as O A - X A Y A Z A, and Y A -axis is coincident with the point A 2. The otions of the end-effect platfor on the botto are caused by the actuators which are posited on the top base 29

40 platfor. The prisatic joints realize the path for the oving legs with the top universal joints. The end-effector platfor oves by taking the reaction fro the botto universal joints which connected to the botto legs of the anipulator [16]. As discussed above, the 3 DOFs represents the translations which are along X axis, Y axis and Z axis since the otions are constrained by the six universal joints (three on the top, three on the botto). In order to analyse the structure of the anipulator, a lib of the entire syste is analyzed. The following figure shows a lib view of the parallel structure hip exoskeleton. C i u i O B b i B i γ i l i P O A a i A i Figure 3.5: A Siplified Lib of the Manipulator Structure As shown in Figure 3.5, the reference coordinate syste O B -X B Y B Z B at the center of the Base-plate is established, Y B -axis will pass through the point B 2. O A -X A Y A Z A at 30

41 the center of the iddle-plate and Y A -axis will pass through the point A 2. Also, we can get the following stateent fro Figure 3.5. B 1. A px py pz T p is the vector fro O B, the origin of the base frae, to O A, the origin of the oving platfor, 2. u u u u i ix iy iz T is the vector fro B i to the first universal joint C i. Note that u i u is the length of the i th actuator, i 3. l l l l i ix iy iz T is the vector fro C i to A i of the i th leg. It should be noted that li is constant for each leg, 4. and define the A and B location (easured fro x) within the horizontal i plane, i i 5. i defines the angle between Ci and the horizontal platfor, 6. defines the nuber of legs, = 1, 2, 3 in this paper. i Dr. Lung-Wen Tsai indicates that a parallel anipulator is classified as syetrical if it satisfies the three conditions which are [2]: The nuber of libs is equal to the nuber of degrees of freedo of the oving platfor; The type and nuber of joints in all the libs are arranged in an identical pattern; The nuber and location of actuated joints in all the libs are the sae. In this thesis, the designed hip exoskeleton has 3DOF structure, and it has three libs. Hence, this parallel anipulator is a syetrical one. 31

42 Parallel echaniss are grouped into three categories in general: planar, spherical, and spatial echaniss. The connectivity of a lib, C k, is a bright criterion to enuerate and classify parallel anipulators by applying the Connectivity Equation [2]. Ck ( 1) M (4) k 1 With a condition of that the connectivity of a lib should not be greater than the otion paraeter and less than the DOF of the oving platfor, which is [2]: M Ck (5) Where: M: is the nubers of degrees of freedo; λ: is the degree of freedo of the space in which a echanis is intended to function; C k : is the syste connectivity. : is the nuber of lib In this thesis, the anipulation has 3 DOF and three libs, so λ =3, M=3. Substituting λ and M into Eq. (4) and (5), we get C1 C2 C3 4M 3 9 (6) And 3 C k 3 (7) 32

43 Therefore, the connectivity of each lib should be equal to 3, which eans that each lib should have 3-DOF in its joints. This anipulator is a planar parallel anipulator because all the oving links in the echanis perfor planar otion [1]. Overall, the P-U-U structure of the hip exoskeleton is feasible based on the above calculation Inverse Kineatics Analysis The purpose of the inverse kineatic analysis is to initialize the oveent of the actuator fro a given position of the obile platfor [17]. As shown in Figure 3.4, the actuator path of prisatic joints is located on the base platfor and the oving platfor (end effector) is parallel to the base, and Z-axis is perpendicular to the base and oving platfors. written as The position vectors of points A i and B i with respect to platfors A and B can be a a cos, a sin,0 T (8) A i i i i i b b cos, b sin,0 T (9) B i i i i i So the actuator vector u i on the base platfor and the end-effector position vector p can be expressed as u u cos cos, u cos sin, u sin (10) i i i i i i i i i T T p p, p, p (11) x y z 33

44 The inverse kineatics analysis is to deterine the displaceent of the actuated variables for a known position and orientation of the end-effector. The joint otions can be derived fro this vector loop as shown in Eq. (12). aicos i px bicos i uicos icos i li ai pi bi ui aisin i py bisin i uicos isin i pz uisin i (12) Which yields: l [ a p b u ] [ a p b u ] 2 T i i i i i i i 2 ( aicos i px bicos i uicos icos i) ( a sin p b sin u cos sin ) ( p u sin ) 2 2 i i y i i i i i z i i (13) The above equation can be written as a function of u i by considering u i is the only unknown. The reason is that the length l i will not change during the operation; i will not change during the operation as well; ai, bi, i, i can be easured; and p is given. To solve u i, Eq. (13) can be written in the for of Eq. (14). u (cos cos cos sin sin ) i i i i i i 2ui cos i cos i( aicos i px bicos i) sin i( aisin i py bisin i) ( pzsin i) ( aicos i px bicos i) ( aisin i py bisin i) pz li 0 (14) rewritten as: Because (cos cos cos sin sin ) 1, the above equation can be i i i i i 34

45 cos ( cos cos ) 2 i ai i px bi i ui 2ui cos i ( pzsin i) sin i( aisin i py bisin i) ( aicos i px bicos i) ( aisin i py bisin i) pz li 0 (15) This is a quadratic function equation of u i. By quadratic rule, there are two solutions for a quadratic equation as the following equation shown with relevant A, B and C. 2 B B 4AC u i (16) 2A However, the solution in this thesis eans the otion of the vector between B i C i, which has to be the position nuber since it eans the actuator is located above the translated leg and enhance the stiffness of the structure [18]. This will be proved in the later section. At this point, there is only one negative solution in this case as shown below regarding to the direction set up for the syste. Also, Since A = 1 in this case, the solution of u i can be defined as 2 B B 4C u i (17) 2 Where: B 2 cos i cos i( aicos i px bicos i) sin i( aisin i py bisin i) ( pzsin i) C ( aicos i px bicos i) ( aisin i py bisin i) pz li (18) 35

46 Since there is one solution for one leg, there are totally three solutions for the given obile platfor position. After each u i is solved, the translation of the entire anipulator is analyzed. A stiffness analysis can be developed fro these results Jacobian Matrix Jacobian atrix is defined as the atrix of all first order partial derivatives of a vector valued function and the otions of this design can be defined by independent paraeters [1]. The Jacobian atrix J of this 3-DOF echanis relates the linear velocity of the oving platfor can be defined by the vector of input joint rates x and the vector of the required actuator otionu. u Jx (19) With T u u1, u2, u3 vx, vy, v x z T (20) In this design, the anipulator has a parallel kineatic structure with intersecting rails. According to Tsai [1], differentiating Eq. (15) with respect to tie yields, x ll u u (21) i i i Where i is the angular velocity of the link A i C i. Dot-ultiplying I i on both sides of Eq. (14) yields l x (22) T l T uu i i i 36

47 Since i=1, 2, 3, Eq. (19) has to be written three ties for the Jacobian atrix for, which gives the final actuator s Jacobian atrix. u Jx (23) With T l1 / l u J l / l u l / l u T 1 1 T T T T (24) Stiffness Analysis There are any eleents which affect the stiffness of a echanis. For a parallel echanis, these eleents are stiffness of the joints, the structure and aterials of the legs and platfors, the geoetry of the structure, and the end-effector position and orientation [3]. In this case, only the wrench applied on the oving platfor will be considered. This wrench can be any force or oent that causes deforations of the oving platfor, which can be described by Eq. (25). w K p (25) Where: K: is the generalized stiffness atrix; w : is the wrench vector eaning the force or the torque that acts on the oving platfor; and it is can be represented as w= Fx Fy Fz x y z ; 37 T

48 p : is the vector of the linear and angular deforation of the oving platfor; and it can be represented as p = x y z x y z T. The wrench vector that is caused by the force and oent applied on the actuators can also be described by the transpose of the Jacobian atrix J, which has been explained by Merlet [12] as the following equations: T w J f (26) q J p (27) Where yields: f : is the vector represents the force or oent of the actuator; q : is the deforation of the actuator. The force/torque and displaceent applied on the actuator can be specified as Eq. (28) by Hookes law. f KJ q (28) Where: K J is the actuator stiffness atrix in a diagonal for that is K J k k k, and k 1, k 2 and k3 represent the joint stiffness of each actuator. Cobined Eqs. (28) and (27) with Eq. (26), the wrench vector becoes: w K p (29) With the stiffness atrix K that can be rewritten as: 38

49 T K J KJJ (30) The results of the stiffness of the hip exoskeleton are shown as in Figures 3.6 to 3.8. Figure 3.6: Stiffness Results in Z direction of the Hip Exoskeleton 39

50 Figure 3.7: Stiffness Results in Y direction of the Hip Exoskeleton Figure 3.8: Stiffness Results in X direction of the Hip Exoskeleton 40

51 First, the above results have certified the syetrical parallel structure that was introduced in Section That is because each diagra shows that stiffness is syetrical referring in all x, y, and z-axes. Second, all three figures show that the results are controlled in a reasonable range, which eans the stiffness of the hip exoskeleton structure is fully tested. Therefore, the design of the anipulator has a reliable structure of the stiffness analysis. 3.4 Sensing and Controlling of the Syste Beside an ideal echanical structure, the sensing and controlling parts of a anipulator design are also critical. Irrational design or selection of the electrical part of a echanis can cause circuit alfunctions, and thus lead to the failure of the entire syste. Therefore, the selection of a powerful sensing and controlling unit is very iportant. The theory of sensing and controlling of the three-degree of freedo parallel structured hip exoskeleton can be siplified in Figure

52 Figure 3.9: Theory of Sensing and Controlling The selected sensors are in the positions where they can detect the certain signals. Then, these signals are transferred to a single chip, which are used for sapling and analyzing. After that, the processed signals are sent to the otors which control the actuator of the echanical. At the end, the end-effector, the lower leg belt achieves the final oveent Sensor Selection and Placeent There are several different types of sensors that can be installed in this syste, such as pressure sensor [19], position sensor [20], and etc. A decision on the selection of 42

53 sensor needs to be ade in further research step. However, circuit alfunction can easily take place if siple or inaccurate sensors are installed to detect an over coplex signal. As a result, the coplexity of the otion signal is one of the factors to consider while selecting sensors for a anipulator. In this case, huan oveent is a otion, and the signals coe fro this otion is also coplex. Though the hip oveent is only a part of the entire huan oveent, it is still considered as a coplex otion. Therefore, advanced and reliable sensors should be chosen for this hip exoskeleton unit in order to avoid circuit alfunction that for ensure the action of the unit. As one of the ost crucial easureent tool approaches for uscle activity, the research of the surface electroyography (SEMG) has been developed significantly in the past ten years [21]. Surface electroyography signals represent the electrical activity of a specific uscle during contractions. As a consequence, SEMG can be used in practical echaniss for a variety of purposes with possible and suitable assistance of odern signal processing ethod, such as the applications in the nerve-uscle-skeleton odeling, artificial lib control, and hip exoskeleton syste [21]. In the design of this hip exoskeleton, the ultichannel surface electroyography sensor is utilized to acquire the SEMG signals that are fro the wearers thigh uscles. The entire signal and data processing circuit is in Figure Ag-Agcl detecting electrodes are stuck to the five priary uscles of thigh, so that the signals collected for SEMG sensors is firstly processed by signal process circuit. After that, the processed SEMG signal is sent to the data processing odule to be analyzed. Then, the 43

54 analyzed SEMG signal is transferred into the standard digital pulse signal, which is used as the coand to control the oveent of otors in further process. Figure 3.10: Surface EMG Signal Processing and Data Processing Circuit According to the earlier chapters, the five priary uscles (Recues feoris, Biceps feoris, Sartorius, Adductor longus, and Adductor agnus) control the hip otion during a noral gait cycle. In other words, Recues feoris and Biceps feoris control the flexion/extension oveent of the thigh; Adductor longus and adductor agnus control the abduction/adduction oveent of the thigh; and Sartorius controls the edial/lateral rotation oveent of the thigh. In this designed unit, the SEMG signals are collected and analyzed by five Ag-AgCl electrodes which are stuck on the surface of the five uscles in order to detect the huan thigh otion during a gait cycle. Due to the properties of the SEMG signals weak and sensitive, soe preparative and additional work needs to be done. For exaple, in order to avoid interruption on the sensor electrodes signal collection, fine hairs need to be reoved and skin needed to be cleansed before sick the electrodes on the proper position of the thigh. 44

55 The choice of locations to place the SEMG sensor electrodes are very iportant because of the need for forbidding the circuit alfunctions. Regarding to the function of the five priary uscles and the properties of the SEMG sensors, the electrodes can be located as the following figures [12]. Figure 3.11: Placeent of SEMG Sensor Electrodes [12] Five electrodes are separately placed on the surface of the thigh skin. In addition, they should be closed to the part of the uscles where there are the axiu activates. Also, these five electrodes are alost evenly disturbed due to the position of uscles, which is a benefit to this design because the influence between electrode signals is iniized [21]. As a result, these five electrodes are connected to the upper leg belt which is on the first working platfor of hip exoskeleton with the best perforance. 45

56 3.4.2 Control Motor Selection and Placeent The selection of the otor could be another ajor issue in this thesis. First, the size of the otor affects the dexterity of the hip exoskeleton. Second, otors have a significant weight which could be the ajor contribution on the whole echanis. The hip exoskeleton is helping the people who cannot walk properly, so the hip exoskeleton does not need a large power assistant that could be harful to the. For this reason, the safety factor of the syste is another issue in the design process which is directly related to the otor. Control is another area needed to be considered with the otors. The efficiency of the otor is also a key point within the design. DC Motor is considered as the first choice in the current research stage. Within this thesis, the otors are considered as an actuator to attain the lower leg belt oveent to assistant oving wearers legs. These otors are located on the waist belt support points as shown in Figure Figure 3.12: The Location of Servo Motors 46

57 Servo otor is first recoendation that can be utilized in the designed unit. The reason is that servo otor is one of the ost coon and accurate otors. There are several advantages by choosing this type of otors. First, Servo otors are well known because its accuracy in controlling the speed of otion and the position of oveent [22]. By using this type of otor, the tolerance of this syste can be iniized and the precision of the oveent can be ensured. Second, the servo otor has a kind of relatively sall device called servo that has an output shaft. This shaft has a fairly fast signal processing tie [23]. In that case, the otor is able to receive the signal and ake the response in a short tie. A siplified servo otor working principle related to the hip exoskeleton is shown in following figure. Figure 3.13: Servo Motor Working Principle To optiize the otor otion, liit switches are adopted. Liit switches are used to control the reaction fro the otors. A basic structure of a liit switch is shown in Figure The theory of the liit switch is that the roller of the liit switch hits the 47

58 lance and cuts out the power of the servo otor to liit the reaction of the upper legs of the exoskeleton when they have reached the expected position [24]. Figure 3.14 [25]: The Structure of Liit Switch: (1): Roller; (2): Upper rocker ar; (3, 5, 11): Spring; (4); Stock; (6): Pulley; (7): Clap; (8, 9) Touch Spot; (10): Transverse plate (25) 3.5 The Hip Exoskeleton Interface to the Wearers In order to take care of the wearers conforability, the design of the anipulator needs to be as adaptable as possible. There are several design issues of this echanis that have been taken into consideration. First, the targeted wearers had either been injured or are weak, so they cannot carry heavy loads. The hip exoskeleton is used to assist their walking during regular daily life; it is ore likely to be a hindrance rather than being helpful if the weight of the echanis is too heavy. In this design, a siple parallel kineatics structure is chosen, 48

59 which has a uch lower oving ass copare to that of serial structure. Second, this design does not require a large aount of building aterials. As a result, this unit can be considered as a lightweight structure when coparing with other present ones. The waist belt of the hip exoskeleton unit is planned to fix on the waist, and the two leg belts are fixed on the thigh. Wearers have different sizes of waist and thighs, and it is ipossible to ake the product custo ade. Therefore, adjustable waist belt and upper leg belt is a potential solution for this proble. In the CAD odel above, the waist belt has been ade in an adjustable for and all the analysis are based on the adjustable design. In the future prototype, both belts should be built in an adjustable for. Due to the structure of this exoskeleton, the wearers are not able to sit while wearing this unit. To solve this proble, this structure has designed to assebly easily. If the upper legs are screwed on the waist belts supporting points, the wearers can take the upper legs out when they need to sit down, and put the back after they stand up. On the other hand, people are able to replace the wear out parts without difficulty with this feature. More than that, there is only one leg exoskeleton (the leg part of the hip exoskeleton unit) attached on the waist belt. If there is a need, two of the can be attached on the waist belts as shown in the following Figure. 49

60 Figure 3.15: CAD Model with two Leg Exoskeletons 50

61 Chapter 4 Syste Optiization The availability and adaptability have been provided by the above analysis, but there is still soething to investigate to optiize the designed unit to achieve a better result on perforance. For instance, further analysis can be done to exaine the kineatic behaviours of the anipulator. In this thesis, the stiffness of the unit is one of the critical conditions for the success of the designed structure. Therefore, the stiffness optiization of the hip exoskeleton is superior suppleentary analysis for verifying and iproving the design. Other iproveents, such as reduction of the aterial used in certain parts (Figure 4.11), can be considered as syste optiization as well. 4.1 FEA Analysis In order to further verify the perforance and kineatic behaviours of the design, finite eleent analysis (FEA) is conducted to perfor the exaination. ANSYS Workbench is used to perfor the tests of the deforation, stress and strain when a force and a oent were applied on this structure. The finite eleent odel is established as shown in Figure 4.1. The three joins (P-U-U) of the unit are taking into consideration during the analysis. The type of the unit is SOLID 187 which is provided by ANSYS Workbench. The eleent is defined by 10 nodes having three degrees of freedo at each node: translations in the nodal X, Y and Z directions. The aterial of the elastic body is 51

62 Structural Steel, and its elastic odulus E = 2.e+011 [Pa], Poisson's ratio μ = 0.3, the density of ρ = 7850 kg Force Figure 4.1: Finite Eleent Model of the Hip Exoskeleton Figure 4.2: Total Deforation based on Applied Force 52

63 The objective of the deforation analysis is to detect the change of the structure when applying a force on the structure. The total deforation shown in Figure 4.2 has been enlarged 1800 ties of its actual deforation when a reasonable size of force is applied on this structure. The force is 200N which is in the direct to the botto of the structure. Fro the figure, we can observe that the first deforing part is the lower leg belt because it is where the force is applied on. Also, the waist belt and the upper legs did not show any deforation even after the result had been enlarged to 1800 ties, which eans the structure of this design is reasonable and reliable. (a) 53

64 (b) (c) Figure 4.3: Deforation in X, Y, Z-axis based on Applied Force: (a) X-axis; (b) Y-axis; (c) Z-axis 54

65 Figure 4.3 shows the coponents of the total deforation in X, Y, and Z-axes, and these three deforation coponents totally different fro each other. Since the force is applied perpendicular to the plane of the lower leg belt, the deforation direction in Z- axis is the largest aong the three directions, and the deforation direction in Y-axis is the sallest. However, all these deforations have been expanded to 1800 ties of its actual size, so these deforations will not daage the structure of the anipulator though there were soe axiu values appeared in the above figures. Figure 4.4: Equivalent Stress based on Applied Force An equivalent stress test (von-mises) also has been done for this structure. Fro the Figure above, the axiu stress in this structure is about 3x10 6 Pa, which is on the lower leg belt and the lower universal joints. However, this result has been enlarged

66 ties. Also, ost parts of this structure are under the iniu stress. Therefore, the equivalent stress analysis is evidenced that the structure of the design is satisfied. Figure 4.5: Equivalent Elastic Strain based on Applied Force The result of the equivalent elastic strain test is very siilar to the result of the equivalent stress test. The axiu strain is on the lower leg belt and the lower universal joints, and ost parts of this structure are under the iniu stain, so the applied force will not destroy the structure of this design. Therefore, this test is also an evidence to prove that this design is coonsensical. 56

67 4.1.2 Moent Figure 4.6: Total Deforation based on Applied Moent The structure of the designed structure has been tested after applying a certain force, and has been proved that the design was reasonable and stable. To verify the above result, a oent is applied on this structure to test the deforation, equivalent stress, and equivalent elastic strain. The sae as the force test, all results has been increased to 1800 ties larger than its original ones. The oent in this case is still applied on the lower leg belt, and perpendicular to the plane of the lower leg belt. The sae as the force test, the axiu deforation is on the botto part of the robot. Soe unexpected rotations appeared in the total deforation Figure, the reason is the transaction had been enlarged to 1800 ties as well. 57

68 It is well known that parallel anipulators have relatively sall workspaces copared with the serial counterparts. (a) (b) 58

69 (c) Figure 4.7: Deforation in X, Y, Z-axis based on Applied Moent: : (a) X-axis; (b) Y-axis; (c) Z-axis Figure 4.7 shows the directional deforations in X, Y, Z-axes when applying a oent on the lower leg belt of the structure, which atched the Figure 4.3. Most parts of the structure are under a relatively sall deforation even the results had been enlarged. The sae as Figure 11 shown, the axiu deforation was in Z-axis and it is , and the sallest deforation was in X-axis that is These deforations are really sall in the tests, and they should be even saller when they are in 1:1 scale. 59

70 Figure 4.8: Equivalent Stress based on Applied Moent Figure 4.9: Equivalent Elastic Strain based on Applied Moent 60

71 Copared Figures 4.8 and 4.9 with Figures 4.4 and 4.5 for the equivalent stress and equivalent elastic strain, they are alost identical except the rotation ones. As discussed above, the rotation can be negligible in a real test. The axiu stress in this structure is about 7x10 7 Pa, which is on the lower leg belt and the lower universal joints. This value is about 20 ties larger than the applied force. Copare with the applied force figure with the applied oent figure, we can obtain that the deforation directions are the sae, but the degrees of deforation are different. The reason is the agnitude of the applied force and the oent were the sae, but the agnitude is not the sae when they are converted to the either force or oent. 4.2 Stiffness Evaluation The workspace of a parallel anipulator can be defined as a reachable region of the origin of a coordinate syste attached to the center of the oving plate [17]. The workspace of the hip exoskeleton can be generated nuerically by MATLAB based on the following algorith: 1. Set volue = 0; 2. Call workspace algorith; 3. Set the steps of θ and φ as θ 0 and φ 0 ; 4. Calculate on unit volue by volue = ρ x ρsinθ x ρsin φ + volue; 5. End of the loop. As a result, the workspace volue is generated as shown in Figure

72 Figure 4.10: Workspace Volue of the Hip Exoskeleton 4.3 Stiffness Optiization In this thesis, the objective of the stiffness optiization is to obtain the best geoetrical paraeters and behaviour paraeters which guide to a better structure of the echanis. In the following optiization process, the Genetic Algoriths (GAs) are used to conduct the optial design of the syste in ters of better syste workspace. GAs is well known to solve both liner and nonlinear probles by presenting a population of possible solutions or individuals, and these solutions/individuals are evaluated with respect to the degree of fitness that indicates how well the solutions/individuals will fit 62

73 the optiization probles [3]. GAs has the following advantage of robustness and good convergence properties [4]: GAs do not require any knowledge or gradient inforation about optiization probles; Discontinuities presents on the optiization probles have little effect on overall operation; Be able to solve large scale probles; Applicable for variety of probles. For the designed hip exoskeleton structure, there are four optiization paraeters chosen for the optiization: the radius of the lower leg belt (R), the radius of the upper leg belt (r), the lower leg length (l), and the angle between the lower leg and the fixed platfor (θ). These paraeters are chosen because they are the variables that have the largest influence of the structure, and these variables aybe have to be adjusted for different set of the hip exoskeleton units. The relationship between the stiffness and these paraeters can be described by the following equation: Kstiffness f( R, r, l, ) (31) With the paraeters potential range which is taken 10% offset fro the original design: 72 R r l 220 o o (32) 63

74 The objective function is convergent to the axiu point after 88 generations as shown in Figure 4.10, and the best choices of the paraeters results that are listed in Table Best: Mean: Best fitness Mean fitness Negtive Stiffness Generation Figure 4.11: Stiffness Optiization Results Table 4.1: Optial Design Paraeters for Maxiizing Stiffness Optial Values Unit 1 st Choice 2 nd Choice 3 rd Choice R r l θ degree K stiffness N/

75 As seen fro the above table, the optial values are alost on the boundary condition. The reason is that the results of nuerical calculation of the optiization have linear relationship. The best result will appear on one end of the range. With the optiized values of the paraeters obtained above, the unit stiffness can be iproved. As a result, the structure of the hip exoskeleton can be iproved, and a better perforance based on stiffness can be achieved. 4.4 Achievable Iproveent of Exoskeleton Developent on Geoetry Other than test the unit and odify paraeters of the design, there are soe other iproveents that can be applied on the structure, such as the developent of the geoetry of the structure (Figure 4.11). 65

76 Figure 4.12: Upgrading CAD Model of the Hip Exoskeleton: (1) Motor (2)Waist Belt (3) Upper leg (4)Sensor connecting point (5) Leg joint (6)Upper leg joint (7)Screw (8) Lower leg(9) Upper leg belt (10) Lower leg joint (11) Lower leg belt As shown in the Figure 4.11, one of the ost obvious changes is the shape of the lower legs. A clearer coparison of the lower legs has been ade in Figure The first design has solid lower legs, but they are flat sticks in the second design. There are two considerations to ake this change. First, the sticks are uch lighter than the solids legs, so it is easier for the wearer. Second, the stick design could save aterial in the anufacturing process, which could reduce the total cost of fabricating the structure. As long as the stiffness is not reduced, the thinner the leg, the better the design is. An FEA analysis of the updated design is presented in the following section to prove that the iproveent of the leg has no negative effect on the stiffness of the structure (Figure 4.13 and Appendix I). 66

77 (A) (B) Figure 4.13: Lower Leg Coparison between Two CAD Models: (A) Updated (B) Original Another iproveent of the odel is the joints. In the original design, the three joints for the legs are in PUU structure, and the three joints ake the 3 DOF of the structure. In the updated design, the prisatic joint stays the sae, but the universal joints are replaced by two revolute joints (Figure 4.13). According to Dr. Lung-Wen Tsai [2], the universal joint is essentially a cobination of two intersecting revolute joints, so the updated odel still has 3DOF [1]. On the other hand, the new joint is uch stronger than the original ones (Figure 4.14 & Appendix I). Thus, the iproveent of the joints only has a positive influence of the perforance of the echanis. 67

78 (A) (B) Figure 4.14: Joints Coparison Between Two CAD Models: (A) Updated (B) Original In order to prove that the changing of the structure actuators does not affect its perforance, a revised FEA has been generated. The results are not exactly the sae as the previous ones, but it is siilar if one copares Figure 4.2 with Figure The axiu deforation is still on the lower leg belt, which eans the changing of joints will not change the stiffness of the entire unit. Consequently, both changes on the new odel are considered as iproveents of the structure. 68

79 Figure 4.15: Total Deforation of Updated Model based on Applied Force More detailed FEA results of the updated CAD odel are in Appendix I and the detailed FEA report is attached in Appendix II. Also, in order to show the hip exoskeleton otion unaffected, detailed otion diagras for the updated CAD odel are provided in the following figures Developent on Cofort Level As entioned above, the hip exoskeleton can perfor translation entions shown in Appendix III. It is not very cofortable to if the oveent of the upper leg is large because there will be angle between the lower platfor and the leg. One solution that can iprove the cofort level of the structure is that to install an inner leg gibal inside of the lower leg belt as shown in Figure

80 Figure 4.16: Lower Leg Belt with an Inner Gibal at Static Position Gibals are considered as a pivoted support that allows the rotation of an object about a single axis [26]. In this design, the inner gibal can be tied on the lower part of the thigh instead of the lower leg belt. At a static position, the lower leg belt and the inner gibal are parallel to each other (see Figure 4.15). When there is a oveent, the lower leg belt still perfors the translation, but the inner gibal rotates with the oveent of the leg as shown in Figure Since this inner gibal is installed inside of the lower leg belt, it does not change the structure of the hip exoskeleton. Therefore, the previous analysis is still available after the inner gibal is installed. The only result is that it increases the cofort level of the entire design. 70

81 Figure 4.17: Lower Leg Belt with an Inner Gibal at Moving Position Due to the sall oveent on the side ways (oving left and right), the inner gibal is good enough for this design. However, a spherical gibal (or called ball joint swivel bearing) can be a better choice for the hip exoskeleton. Inside of the lower leg belt, there is a ball shell joint as shown in Figure Figure 4.18: Lower Leg Belt with a Spherical Gibal at Static Position 71

82 (a) (b) Figure 4.19: Lower Leg Belt with a Spherical Gibal at Moving Position: (a) Forward; (b) Sideways 72

83 The advantage of a spherical gibal is that it can rotate in all directions. Therefore, it avoids the liitation of the inner gibal (copare Figure 4.18 with Figure 4.16). Another advantage of the spherical gibal is that there is no screw needed in this unit, so it is ore cofortable for the wearers. The sae as the inner gibal, the spherical gibal does not change the structure of the hip exoskeleton. As a result, the previous analysis is still available. 73

84 Chapter 5 Results and Discussion With the theoretical results of the analysis, the CAD odel is able to perfor as required as shown in the following figures (Figure ). Figure 5.1: Move Forward Figure 5.2: Move Backward 74

85 Figure 5.3: Move Right Figure 5.4: Move Left Based on the previous analysis and the perforance, the CAD odel shows the viability of the parallel structure of the hip exoskeleton unit, which introduced the idea of partial anipulator. This echanis is siple, lightweight, and has no extra coponents, and it should contain all the advantages that a parallel echanis has. The obility analysis ade in section shows that this unit is able to perfor the three-degree of freedo that a noral hip otion needs. The inverse kineatics 75

86 analysis presents the basic otion needed fro the syste by calculating the displaceent of the actuated variables. By using this result of Jacobian Matrix, the first order partial derivatives of a vector valued function and the otions of this unit are defined by independent paraeters. Until this point, the parallel configuration has been copletely evaluated and been authenticated that this syste is adaptable. The FEA analysis verified the perforance and stiffness of the syste by providing very siilar results with these ones calculated in MatLab. Though the stiffness of the syste is not the only condition for aking the structure, it offers a further iproveent on the kineatic properties and structure of the design. Consequently, the optiization of the stiffness has proved that the structure of the designed echanis is in a reliable range. These above results show that this thesis has fully answered the hypothesis proposed in Chapter 1. It is reasonable to design a three-degree of freedo parallel structure hip exoskeleton unit that helps wearers to ove their legs properly, and the parallel structure designed for this anipulator is novel reliable, convenient, and state-ofthe-art. Acceding to these results obtained, there are still soe iproveents which can be applied on this unit, such as a shoulder harness. With a shoulder harness, the weight of the hip exoskeleton is carried by the shoulders instead of the waist. As a result, the wears feel uch ore cofortable. 76

87 Chapter 6 Conclusions In this thesis, the design of a three-degree of freedo parallel structure hip exoskeleton unit has been described in detail. While designing this structure, a list of requireents on structure, applicability, and econoic consideration extracted fro the noral huan hip otion analysis is kept included. This hip exoskeleton is a novel one: first, the hip exoskeleton is fully built with a parallel structure, so that the advantages of a parallel structure are also owned, such as higher accuracy and stiffness, lower oving ass, and reduced installation requireents [4]. Also, coparing with the existing units, it is uch saller, which eans it is uch lighter and convenient to the. Last but not least, the P-U-U configuration of the exoskeleton unit eets all the needs of a hip otion. A series of analysis of the structure has been copleted in this thesis. - In order to achieve the basic hip activity, the exoskeleton unit has to accoodate the hip s threedegree of freedo otions which are flexion/extension, abduction/adduction, and edial/lateral rotation. The obility analysis in the previous section has been proved that the P-U-U configuration eets this condition. The calculated inverse kineatic analysis and the Jacobian atrix analysis proved and supports that the parallel structure is reliable and accurate. The stiffness analysis proved that the structure of the anipulator is valid and reasonable. FEA analysis of the unit is another strong support of this design. Optiization that is used to analyze the requireents on stiffness, the echanical 77

88 interferences and the geoetric properties have been provided as well. Other than verify the structure of the hip exoskeleton, an additional work on sensing and control has been discussed in this thesis, which akes the future work of the entire project uch easier. Overall, this research has achieved the objectives of the thesis, and it is reasonable to design a three-degree of freedo parallel structure hip exoskeleton unit that helps wearers oving their legs properly. At the sae tie, the designed structure is reliable, convenient, and state-of-the-art. In addition to augenting the coplexity of echanis, the advanceent of this hip exoskeleton unit can contribute to the science of partial anipulator and lead to a better understanding of locootory bioechanics [13] in an econoical view in the future. 78

89 Chapter 7 Future Work The structure has been tested and it has been approved that this design is reasonable and attainable, which eans it is ready to go further. Therefore, the following fields can be considered as future work for the hip exoskeleton: building a prototype, chose the aterials and perfor further optiization on the structure of the prototype. 7.1 Prototype Currently, the hip exoskeleton has been built in SolidWorks as a CAD odel. Though the anipulator has satisfied all the visual ode requireents, a prototype will help to solve all the potential probles. Also, a prototype has uch ore contributions to a product than a CAD odel does. First of all, prototype is a real odel of the anipulator, so it provides a visual odel. Based on the visual odel, soe adjustents can be easily ade, such as the size of the waist belt. These sall adjustents will not affect the perforance of the hip exoskeleton, but it will iprove the applicability. It is hard to ake these kind adjustents on a coputer odel than a prototype due to the intangibility of the CAD odel. Second, the hip exoskeleton odel is oving perfectly in the progra as shown in the previous chapters; however, we cannot assue that the wearers will feel perfectly 79

90 cofortable until they wear the actual hip exoskeleton on the. After the wearers have the anipulator on, they can find out how they feel and where need to be iproved. Coparing with a coputer odel, it is easier to optiize the entire design on a prototype because it is easier to be adjusted. Third, there are no errors or tolerances in a CAD odel since every detention is ideal, but it is different in a prototype. Errors and tolerances are unavoidable in the real world, so a prototype is uch closer to the reality than a coputer odel. As a result, a prototype provides uch accurate data than a coputer odel for the future. Last, a prototype can open people s ind better. After wearing a prototype, the designer will be able to understand it better and have ore ideas on iproving. 7.2 Materials As entioned above, this hip exoskeleton anipulator is build for the people who have difficulty in oving their hips. As a consequence, ost of those people are not able to carry heavy load, such as a heavy hip exoskeleton. Therefore, the lighter the anipulator is, the ore the wearers prefer, so the selection of the aterial of the hip exoskeleton can be another ajor subject in this case. In order to keep the properties of this design, a strong, light, and tough aterial should be chosen. However, soe light and suitable aterials can be extreely expensive as anufacturing aterials, such as, titaniu, the aterial to build space station [25]. This design is targeted to average revenue faily; there is no point to ake this product expensive as ost will not be able to afford it if is. As a result, how to 80

91 balance the design properties and aterial cost should be another consideration keep in ind when selecting the aterials. As suggestion, aluinu alloy is a first-rate choice since it is a tough, strong, and lightweight aterial [26]. Coparing with titaniu, it is uch cheaper. By choosing this aterial, the properties of this design can be fully kept. Certainly, there are soe other aterials to choose by doing further research, but this step should be able to coplete in the prototype stage in order to save both oney and tie. 7.3 Sensor and Motor The ajor task for this thesis is to build a parallel structure hip exoskeleton and authenticate the structure by ultiple ethods, which has been copleted within the paper. This thesis also provides brief suggestions about the sensors and otors that should be chosen for this novel design. The criteria for sensors and otors have been specified in the previous chapter, and further researches need to be working on to select specific odel to build the anipulator prototype in the future. 7.4 Further Test and Optiization After building the prototype, soe adjustents and iproveents can be ade on the actual anipulator. Most of potential changes should not affect the structure of this design to keep the basic perforance, but few of the should be able to iprove the quality of the hip exoskeleton unit. In order to get ore accurate data for the future production, soe tests and analysis on both the coputer odel and prototype have to be re-done after all the odifications. 81

92 Not all of the parts of this anipulator will have the sae length of usage duration, which eans soe parts will break before other parts. In order to iniu the cost, the wearers will prefer to change the broken parts rather than purchase a new unit. Also, soe wearers ay not need to go to outside either due to their health conditions or the weather conditions. Nevertheless, they do not want to take the unit off because they still need to walk around. In these cases, the anipulator has to be easily assebled and disassebled as entioned in the previous chapter. In order to best satisfy this need, the optiization of the asseblage should be achieved in the future as well. Accessional ites can be added on this unit in consideration of suppleentary requireents fro custoers. Soe of the wearers ay not be able to put the unit on because their waists cannot carry the heavy load though the anipulator is fairly light. In that case, a pair of suspenders will be the best solution, which eans the entire weight of the unit is on the shoulders rather than on the waist. This kind of optiization can always iprove the range of the use of the anipulator. Overall, there is still soe roo for iproveent for the hip exoskeleton unit in order to lead an ideal position in the arket after it has been produced. 82

93 Reference [1] Tsai, Lung-Wen. Robot Analysis. New York : John Wiley & Sons. Inc, [2] Pratt, Jerry E. The RoboKnee: An Exoskeleton for Enhancing Strength and Endurance During Walking. New Orleans : IEEE, July 2004, Vol. 3. pp [3] Pan, Min; Chi, Zhongzhe and Zhang, Dan. Novel Design of a Three DOFs MEMSbased Precision Manipulator. Beijing : IEEE International Conference on Mechatronics and Autoation, [4] Zhang, Dan. Parallel Robotic Machine Tools. New York : Springer Science+Business Media, [5] Robots for All. [Online] Septeber 26, [Cited: October 16, 2010.] X9dZGI/AAAAAAAAACI/wdMfAjCSN84/s400/adept.png&igrefurl= ics.blogspot.co/. [6] Huston, Ronald L. Principles of Bioechanics. Boca Raton : Taylor & Francis Group, : [7] Valiente, Andrew. Design of a Quasi-Passive Parallel Leg Exoskeleton to Augent Load Carrying for Walking. Cabridge : LeTourneau University, [8] Cyberdyne. Robot Suit Hal. [Online] Cyberdyne, Inc, Auguest [Cited: Noverber 2, 2010.] [9] Kazerooni, Racine and Huang, Steger. On the Control of the Berkeley Lower Extreity Exoskeleton (BLEEX). Barcelona : IEEE, April International Conference on Robotics and Autoation. pp [10] Haill, Joshph and Knutzen, Kathleen M. Bioechanical: Basis of Huan Moveent. Baltiore : Lippincott Willias & Wikins, [11] Masood, M.A. Techhilla. Covers Apple, Microsoft, Google & the Web. [Online] August 9, [Cited: May 25, 2011.] [12] Perry, Jacqelin and Burnfield, Judith M. Gait Analysis: Noral and Pathological Function. New Jersey : SLACK Incorporated,

94 [13] Walsh, Jaes. Bioietic Design of an Under-Actuated Leg Exoskeleton For Load- Carrying Augentation. Cabbrege : Massachusetts Institute of Technology, [14] Oatis, Carol A. Kinesiology: the Mechanics & Pathoechanics of Huan Moveent. Baltiore : Lippincott Willias & Wilkins, [15] Tsai, Lung-Wen and Saeer, Joshi. Kineatic analysis of 3-DOF position echaniss for use in hybrid kineatic achine. Journal of Mechanical Design, June 2002, Journal of Mechanical Design, Vol. 124, pp [16] Masory, Oren and Wang, Jian. Workspace evauation of stewart platfors. 1995, Advanced Robotics, Vol 9, No.4, pp [17] Xu, Qingsong and Li, Yangin. Kineatic analysis and design of a new 3-DOF translational parallel anipulator. Jounral of Intellgent and Robotic Systes, July 2006, Journal of Mechanical Design, Vol.128, pp [18] Li, Yangin and Xu, Qingsong. A new approach to the architecture optiization of a general 3-PUU translational parallel anipulator. July 2006, Journal of Intelligent and Robotic Systes, Vol.46, pp [19] Meyer, Jan; Lukowicz, Paul and Troster, Gerhard. Textile Pressure Sensor for Muscle Activity and Motion Dection. Montreux : Wearable Coputers (ISWC), October, 2006, Vol.6. pp [20] Herens, Herie J.; Freriks, Bart and et al. Developent of Recoendations for SEMG Sensors and Senor Placeent Procedures. Journal of Electroyography and Kinesiology, 2000, Vol. 2. pp [21] Gao, Zhen and Zhang, Dan. Surface Electroyographic Sensor for Huan Motion Estiation Based on Ar Wrestling Robot. Sensors & Transducers Journal, June, 2010, Vol. 117, pp [22] Norton, Rober L. Machine Design: An Integrated Approach. Upper Saddle River : Pearson Education, Inc., [23] Standard Technologies of the Seattel Robotics Society. What is a Servos. [Online] Seattle Robotics Society, July [Cited: May 26, 2011.] [24] China Motor Inforation. DZJ. [Online] Nibe Xing Huang Autoobile Service Ltd., May 09, [Cited: May 27, 2011.] 84

95 [25] Callister, Willia D. Jr. Materials Science and Engineering: An Introduction. New York : John Wiley & Sons, Inc., [26] Bedrossian, Nazareth S.; Paradiso, Joseph and et al. Steering law design for redundant single-gibal control oent gyroscopes. Journal of Guidance, Control and Synaics, 1990, Vol. 13, pp

96 Appendix Appendix I: Iproved Structure FEA Analysis Figure AI.1: Finite Eleent Model of the Updating Hip Exoskeleton Appendix I.I: Force Application Figure AI.2: Total Deforation based on Applied Force 86

97 Figure AI.3: Deforation in X, Y, Z-axis based on Applied Force Figure AI.4: Equivalent Stress based on Applied Force 87

98 Figure AI.5: Equivalent Strain based on Applied Force Appendix I.II: Moent Application Figure AI.6: Total Deforation based on Applied Moent 88

99 Figure AI.7: Deforation in X, Y, Z-axis based on Applied Moent Figure AI.8: Equivalent Stress based on Applied Moent 89

100 Figure AI.9: Equivalent Strain based on Applied Moent 90

101 Appendix II: FEA Report of the Iproved Structure Project First Saved Monday, May 30, 2011 Last Saved Monday, May 30, 2011 Product Version Release 91

102 Contents Units Model (B4) o Geoetry Parts o Coordinate Systes o Connections Contact Regions o o Mesh Static Structural (B5) Analysis Settings Loads Solution (B6) Solution Inforation Results Material Data o Structural Steel Units TABLE 1 Unit Syste Metric (, kg, N, s, V, A) Degrees rad/s Celsius Angle Rotational Velocity Teperature Degrees rad/s Celsius Model (B4) Geoetry Object Nae State TABLE 2 Model (B4) > Geoetry Geoetry Fully Defined Definition Source Type H:\Users\Adinistrator\AppData\Local\Tep\WB_WIN- 388KFD85MH7_5548_2\unsaved_project_files\dp0\SYS-1\DM\SYS-1.agdb DesignModeler 92

103 Length Unit Eleent Control Display Style Meters Progra Controlled Part Color Bounding Box Length X Length Y Length Z Properties Volue e-004 ³ Mass kg Scale Factor Value 1. Statistics Bodies 32 Active Bodies 32 Nodes Eleents Mesh Metric None Preferences Iport Solid Bodies Iport Surface Bodies Iport Line Bodies Paraeter Processing Personal Paraeter Key Yes Yes No Yes DS 93

104 CAD Attribute Transfer Naed Selection Processing Material Properties Transfer CAD Associativity Iport Coordinate Systes Reader Save Part File Iport Using Instances Do Sart Update Attach File Via Tep File Teporary Directory No No No Yes No No Yes No Yes H:\Users\Adinistrator\AppData\Local\Tep Analysis Type 3-D Mixed Iport Resolution Enclosure and Syetry Processing None Yes TABLE 3 Model (B4) > Geoetry > Parts Upper Leg Object Nae Leg Joint-2 Leg Joint-3 Joint-1 Upper Leg Joint-2 Upper Leg Joint-3 State Meshed Graphics Properties Visible Yes Transparency 1 94

105 Definition Suppressed Stiffness Behavior Coordinate Syste Reference Teperature No Flexible Default Coordinate Syste By Environent Material Assignent Nonlinear Effects Theral Strain Effects Structural Steel Yes Yes Bounding Box Length X e e e-002 Length Y e e-002 Length Z e e e-002 Properties Volue e-005 ³ e-006 ³ Mass kg e-002 kg 7.803e-002 kg Centroid X e e e e Centroid Y e e e e e-002 Centroid Z e-003 Moent of Inertia Ip e-005 kg ² e-005 kg ² e-005 kg ² Moent of Inertia Ip e-005 kg ² e-006 kg ² 3.531e-006 kg ² e-006 kg ² Moent of Inertia e e e-005 kg ² e

106 Ip3 kg ² kg ² kg ² Statistics Nodes Eleents Mesh Metric None TABLE 4 Model (B4) > Geoetry > Parts Object Nae Scaw-1 Scaw-2 Scaw-3 Scaw-4 Scaw-5 State Meshed Graphics Properties Visible Yes Transparency 1 Definition Suppressed Stiffness Behavior Coordinate Syste Reference Teperature No Flexible Default Coordinate Syste By Environent Material Assignent Nonlinear Effects Theral Strain Effects Structural Steel Yes Yes Bounding Box Length X e e e e e-002 Length Y e e e e e

107 Length Z e e e e e-002 Properties Volue e-006 ³ Mass e-002 kg Centroid X e e e e- 002 Centroid Y e e e e e- 002 Centroid Z e e e- 002 Moent of Inertia Ip e-007 kg ² e-006 kg ² 3.382e-007 kg ² e-007 kg ² Moent of Inertia Ip e-006 kg ² e-006 kg ² e-006 kg ² Moent of Inertia Ip e-006 kg ² e-007 kg ² e-006 kg ² Statistics Nodes Eleents Mesh Metric None Object Nae TABLE 5 Model (B4) > Geoetry > Parts Upper Leg Upper Leg Scaw-6 Joint-4 Joint-5 Upper Leg Joint-6 Lower Leg Belt-3 State Meshed Graphics Properties Visible Yes Transparency 1 Definition 97

108 Suppressed Stiffness Behavior Coordinate Syste Reference Teperature No Flexible Default Coordinate Syste By Environent Material Assignent Nonlinear Effects Theral Strain Effects Structural Steel Yes Yes Bounding Box Length X e e Length Y e e e-002 Length Z 4.e e Properties Volue e-006 ³ e-006 ³ e-005 ³ Mass e-002 kg e-002 kg kg Centroid X e e-002 Centroid Y e Centroid Z e e e e e-003 Moent of Inertia Ip e-006 kg ² e-005 kg ² e-005 kg ² e-003 kg ² Moent of Inertia Ip e-006 kg ² e-006 kg ² e-006 kg ² e-003 kg ² 98

109 Moent of Inertia Ip e-007 kg ² e-005 kg ² e-005 kg ² e-003 kg ² Statistics Nodes Eleents Mesh Metric None TABLE 6 Model (B4) > Geoetry > Parts Object Nae Sensor-1 Sensor-2 Sensor-3 Upper Leg Belt-2 leg-1 State Meshed Graphics Properties Visible Yes Transparency 1 Definition Suppressed Stiffness Behavior Coordinate Syste Reference Teperature No Flexible Default Coordinate Syste By Environent Material Assignent Nonlinear Effects Theral Strain Effects Structural Steel Yes Yes Bounding Box Length X e e e e

110 Length Y 1.6e e Length Z 1.666e e e e-002 Properties Volue e-006 ³ e-005 ³ e-006 ³ Mass e-002 kg kg e-002 kg Centroid X e e e e-002 Centroid Y e e e e e-002 Centroid Z e e e e Moent of Inertia Ip e-007 kg ² e-007 kg ² e-007 kg ² e-003 kg ² e-005 kg ² Moent of Inertia Ip e-007 kg ² e-007 kg ² e-007 kg ² e-003 kg ² e-007 kg ² Moent of Inertia Ip e-007 kg ² e-007 kg ² e-007 kg ² e-003 kg ² e-005 kg ² Statistics Nodes Eleents Mesh Metric None TABLE 7 Model (B4) > Geoetry > Parts Object Nae leg-2 leg-3 leg-4 leg-7 leg-8 State Meshed Graphics Properties Visible Yes 100

111 Transparency 1 Definition Suppressed Stiffness Behavior Coordinate Syste Reference Teperature No Flexible Default Coordinate Syste By Environent Material Assignent Nonlinear Effects Theral Strain Effects Structural Steel Yes Yes Bounding Box Length X e e e-002 Length Y Length Z e e e-002 Properties Volue e-006 ³ Mass e-002 kg Centroid X e e e Centroid Y e e e e e-002 Centroid Z e e e-002 Moent of Inertia Ip e-005 kg ² e-005 kg ² 5.647e-005 kg ² e-005 kg ² e-005 kg ² Moent of Inertia e e e e e

112 Ip2 kg ² kg ² kg ² kg ² kg ² Moent of Inertia Ip e-005 kg ² e-005 kg ² e-005 kg ² e-005 kg ² e-005 kg ² Statistics Nodes Eleents Mesh Metric None Object Nae Upper Leg-4 TABLE 8 Model (B4) > Geoetry > Parts Upper Leg-5 Upper Leg-6 Boss- Extrude3[1] Boss- Extrude3[2] State Meshed Graphics Properties Visible Yes Transparency 1 Definition Suppressed Stiffness Behavior Coordinate Syste Reference Teperature No Flexible Default Coordinate Syste By Environent Material Assignent Nonlinear Effects Theral Strain Effects Structural Steel Yes Yes Bounding Box Length X e e e e

113 Length Y e-002 Length Z 3.e e e e-002 Properties Volue e-005 ³ e-005 ³ Mass kg kg Centroid X e e e e-002 Centroid Y Centroid Z e e Moent of Inertia Ip e-003 kg ² e-003 kg ² e-006 kg ² e-006 kg ² Moent of Inertia Ip e-005 kg ² e-005 kg ² e-005 kg ² Moent of Inertia Ip e-003 kg ² e-003 kg ² e-005 kg ² Statistics Nodes Eleents Mesh Metric None TABLE 9 Model (B4) > Geoetry > Parts Object Nae Chafer2 Leg Joint-1 State Meshed Graphics Properties Visible Yes Transparency 1 103

114 Definition Suppressed Stiffness Behavior Coordinate Syste Reference Teperature No Flexible Default Coordinate Syste By Environent Material Assignent Nonlinear Effects Theral Strain Effects Structural Steel Yes Yes Bounding Box Length X e-002 Length Y 2.4e e-002 Length Z e-002 Properties Volue e-004 ³ e-005 ³ Mass kg kg Centroid X e Centroid Y e-002 Centroid Z e e-003 Moent of Inertia Ip e-003 kg ² e-005 kg ² Moent of Inertia Ip e-002 kg ² e-005 kg ² Moent of Inertia Ip e-002 kg ² e-005 kg ² Statistics Nodes Eleents

115 Mesh Metric None Coordinate Systes TABLE 10 Model (B4) > Coordinate Systes > Coordinate Syste Object Nae Global Coordinate Syste State Fully Defined Definition Type Cartesian Ansys Syste Nuber 0. Origin Origin X Origin Y Origin Z Directional Vectors X Axis Data [ ] Y Axis Data [ ] Z Axis Data [ ] Connections TABLE 11 Model (B4) > Connections Object Nae Connections State Fully Defined Auto Detection Generate Contact On Update Tolerance Type Yes Slider Tolerance Slider

116 Tolerance Value e-003 Face/Face Face/Edge Edge/Edge Priority Group By Search Across Revolute Joints Fixed Joints Yes No No Include All Bodies Bodies Yes Yes Transparency Enabled Yes Object Nae State Scoping Method TABLE 12 Model (B4) > Connections > Contact Regions Contact Contact Region 2 Region 3 Contact Region Scope Fully Defined Geoetry Selection Contact Region 4 Contact Region 5 Contact 2 Faces 9 Faces 2 Faces 9 Faces 1 Face Target 2 Faces 9 Faces 2 Faces 9 Faces 1 Face Contact Bodies Leg Joint-2 Leg Joint-3 Upper Leg Joint-1 Target Bodies Upper Leg Joint-2 Upper Leg-6 Upper Leg Joint-1 Upper Leg-5 Scaw-3 Definition Type Scope Mode Bonded Autoatic 106

117 Behavior Suppressed Syetric No Advanced Forulation Noral Stiffness Update Stiffness Pinball Region Pure Penalty Progra Controlled Never Progra Controlled Object Nae State Scoping Method TABLE 13 Model (B4) > Connections > Contact Regions Contact Contact Region 7 Region 8 Contact Region 6 Scope Fully Defined Geoetry Selection Contact Region 9 Contact Region 10 Contact 1 Face 2 Faces 1 Face Target 1 Face 2 Faces 1 Face Contact Bodies Upper Leg Joint-2 Upper Leg Joint-3 Scaw-1 Target Bodies Scaw-1 Scaw-2 Leg Joint-1 leg-3 leg-4 Definition Type Scope Mode Behavior Suppressed Bonded Autoatic Syetric No Advanced 107

118 Forulation Noral Stiffness Update Stiffness Pinball Region Pure Penalty Progra Controlled Never Progra Controlled Object Nae State Scoping Method Contact Target TABLE 14 Model (B4) > Connections > Contact Regions Contact Contact Region 12 Region 13 Contact Region 11 Scope Fully Defined Geoetry Selection 1 Face 1 Face Contact Region 14 Contact Region 15 Contact Bodies Scaw-2 Scaw-3 Scaw-4 Target Bodies leg-7 leg-8 leg-1 leg-2 Upper Leg Joint-5 Definition Type Scope Mode Behavior Suppressed Bonded Autoatic Syetric No Advanced Forulation Noral Stiffness Pure Penalty Progra Controlled 108

119 Update Stiffness Pinball Region Never Progra Controlled Object Nae State Scoping Method Contact Target TABLE 15 Model (B4) > Connections > Contact Regions Contact Contact Region 17 Region 18 Contact Region 16 Scope Fully Defined Geoetry Selection 1 Face 1 Face Contact Region 19 Contact Region 20 Contact Bodies Scaw-4 Scaw-5 Target Bodies leg-3 leg-4 Upper Leg Joint-6 leg-1 leg-2 Definition Type Bonded Scope Mode Autoatic Behavior Syetric Suppressed No Advanced Forulation Pure Penalty Noral Stiffness Progra Controlled Update Stiffness Never Pinball Region Progra Controlled 109

120 Object Nae State Scoping Method Contact Target TABLE 16 Model (B4) > Connections > Contact Regions Contact Contact Region 22 Region 23 Contact Region 21 Scope Fully Defined Geoetry Selection 1 Face 1 Face Contact Region 24 Contact Region 25 Contact Bodies Scaw-6 Upper Leg Joint-4 Upper Leg Joint-5 Target Bodies Type Scope Mode Behavior Suppressed Forulation Noral Stiffness Update Stiffness Pinball Region Upper Leg Joint-4 leg-7 leg-8 Lower Leg Belt-3 Definition Bonded Autoatic Syetric No Advanced Pure Penalty Progra Controlled Never Progra Controlled Object Nae TABLE 17 Model (B4) > Connections > Contact Regions Contact Contact Region 27 Region 28 Contact Region 26 Contact Region 29 Contact Region

121 State Fully Defined Scope Scoping Method Geoetry Selection Contact 1 Face 2 Faces 1 Face Target 1 Face 2 Faces 1 Face Contact Bodies Upper Leg Joint-6 Sensor-1 Sensor-2 Sensor-3 Upper Leg Belt-2 Target Bodies Lower Leg Belt-3 Upper Leg Belt-2 Upper Leg-4 Definition Type Bonded Scope Mode Autoatic Behavior Syetric Suppressed No Advanced Forulation Pure Penalty Noral Stiffness Progra Controlled Update Stiffness Never Pinball Region Progra Controlled Object Nae State Scoping Contact Region 31 TABLE 18 Model (B4) > Connections > Contact Regions Contact Contact Region Region Scope Fully Defined Geoetry Selection Contact Region 34 Contact Region

122 Method Contact 1 Face 9 Faces 1 Face Target 1 Face 9 Faces 1 Face Contact Bodies Upper Leg Belt-2 Upper Leg-4 Upper Leg-5 Target Bodies Upper Leg-5 Upper Leg-6 Chafer2 Leg Joint-1 Boss- Extrude3[1] Definition Type Scope Mode Behavior Suppressed Bonded Autoatic Syetric No Advanced Forulation Noral Stiffness Update Stiffness Pinball Region Pure Penalty Progra Controlled Never Progra Controlled Object Nae State Scoping Method Contact Target Contact Region 36 TABLE 19 Model (B4) > Connections > Contact Regions Contact Region Contact 37 Region 38 Scope Fully Defined Geoetry Selection 1 Face 1 Face Contact Region 39 Contact Region

123 Contact Bodies Upper Leg-5 Upper Leg-6 Boss- Extrude3[1] Boss- Extrude3[2] Target Bodies Chafer2 Boss- Extrude3[2] Chafer2 Definition Type Bonded Scope Mode Autoatic Behavior Syetric Suppressed No Advanced Forulation Pure Penalty Noral Stiffness Update Stiffness Pinball Region Progra Controlled Never Progra Controlled Mesh TABLE 20 Model (B4) > Mesh Object Nae State Mesh Solved Defaults Physics Preference Mechanical Relevance 0 Sizing Use Advanced Size Function Relevance Center Eleent Size Off Coarse Default 113

124 Initial Size Seed Soothing Transition Span Angle Center Miniu Edge Length Active Assebly Mediu Fast Coarse e-004 Inflation Use Autoatic Tet Inflation Inflation Option None Sooth Transition Transition Ratio Maxiu Layers 5 Growth Rate 1.2 Inflation Algorith View Advanced Options Pre No Advanced Shape Checking Eleent Midside Nodes Straight Sided Eleents Standard Mechanical Progra Controlled No Nuber of Retries Default (4) Rigid Body Behavior Diensionally Reduced Mesh Morphing Disabled Pinch Pinch Tolerance Generate on Refresh Please Define No Statistics Nodes

125 Eleents Mesh Metric None Static Structural (B5) TABLE 21 Model (B4) > Analysis Object Nae Static Structural (B5) State Solved Definition Physics Type Analysis Type Structural Static Structural Solver Target ANSYS Mechanical Options Environent Teperature 22. C Generate Input Only No TABLE 22 Model (B4) > Static Structural (B5) > Analysis Settings Object Nae Analysis Settings State Fully Defined Step Controls Nuber Of Steps 1. Current Step Nuber Step End Tie Auto Tie Stepping s Progra Controlled Solver Controls Solver Type Progra Controlled 115

126 Weak Springs Large Deflection Inertia Relief Progra Controlled Off Off Nonlinear Controls Force Convergence Moent Convergence Displaceent Convergence Rotation Convergence Line Search Progra Controlled Progra Controlled Progra Controlled Progra Controlled Progra Controlled Output Controls Calculate Stress Calculate Strain Calculate Results At Yes Yes All Tie Points Analysis Data Manageent Solver Files Directory Future Analysis H:\Users\Adinistrator\Desktop\in\Graduate Study\Thesis\ \ Ansys\New Design FEA_files\dp0\SYS-1\MECH\ None Scratch Solver Files Directory Save ANSYS db Delete Unneeded Files Nonlinear Solution Solver Units No Yes No Active Syste 116

127 Solver Unit Syste ks TABLE 23 Model (B4) > Static Structural (B5) > Loads Object Nae Fixed Support Force State Fully Defined Scope Scoping Method Geoetry Geoetry Selection 1 Face Definition Type Fixed Support Force Suppressed No Define By Vector Magnitude 100. N (raped) Direction Defined FIGURE 1 Model (B4) > Static Structural (B5) > Force 117

128 Solution (B6) TABLE 24 Model (B4) > Static Structural (B5) > Solution Object Nae Solution (B6) State Solved Adaptive Mesh Refineent Max Refineent Loops 1. Refineent Depth 2. TABLE 25 Model (B4) > Static Structural (B5) > Solution (B6) > Solution Inforation Object Nae Solution Inforation State Solved Solution Inforation Solution Output Solver Output Newton-Raphson Residuals 0 Update Interval Display Points 2.5 s All Object Nae State Scoping Method Geoetry TABLE 26 Model (B4) > Static Structural (B5) > Solution (B6) > Results Total Directional Directional Directional Deforation Deforation Deforation 2 Deforation 3 Scope Solved Geoetry Selection Definition All Bodies Equivalent Elastic Strain Type Total Deforation Directional Deforation Equivalent (von- Mises) Elastic 118

129 Strain By Display Tie Calculate Tie History Tie Last Yes Identifier Orientation X Axis Y Axis Z Axis Coordinate Syste Global Coordinate Syste Use Average Yes Results Miniu e e e e-014 / Maxiu e e e e e-005 / Miniu Occurs On Chafer2 leg-7 Lower Leg Belt-3 leg-4 Chafer2 Maxiu Occurs On leg-8 leg-3 leg-7 leg-1 Lower Leg Belt-3 Inforation Tie 1. s Load Step 1 Substep 1 Iteration Nuber 1 TABLE 27 Model (B4) > Static Structural (B5) > Solution (B6) > Results Object Nae Equivalent Stress State Solved 119

130 Scope Scoping Method Geoetry Geoetry Selection All Bodies Definition Type Equivalent (von-mises) Stress By Display Tie Calculate Tie History Use Average Tie Last Yes Yes Identifier Results Miniu Maxiu Miniu Occurs On Maxiu Occurs On 4.828e-003 Pa e+006 Pa Chafer2 Lower Leg Belt-3 Inforation Tie 1. s Load Step 1 Substep 1 Iteration Nuber 1 Material Data Structural Steel TABLE 28 Structural Steel > Constants Density 7850 kg ^-3 Coefficient of Theral Expansion 1.2e-005 C^-1 120

131 Specific Heat 434 J kg^-1 C^-1 Theral Conductivity 60.5 W ^-1 C^-1 Resistivity 1.7e-007 oh TABLE 29 Structural Steel > Copressive Ultiate Strength Copressive Ultiate Strength Pa 0 TABLE 30 Structural Steel > Copressive Yield Strength Copressive Yield Strength Pa 2.5e+008 TABLE 31 Structural Steel > Tensile Yield Strength Tensile Yield Strength Pa 2.5e+008 TABLE 32 Structural Steel > Tensile Ultiate Strength Tensile Ultiate Strength Pa 4.6e+008 TABLE 33 Structural Steel > Alternating Stress Alternating Stress Pa Cycles Mean Stress Pa 3.999e e e e e e

132 2.62e e e e e e e e Strength Coefficient Pa Strength Exponent TABLE 34 Structural Steel > Strain-Life Paraeters Ductility Coefficient Ductility Exponent Cyclic Strength Coefficient Pa Cyclic Strain Hardening Exponent 9.2e e TABLE 35 Structural Steel > Relative Pereability Relative Pereability TABLE 36 Structural Steel > Isotropic Elasticity Teperature C Young's Modulus Pa Poisson's Ratio 2.e