A Laboratory Demonstration of a Parallel Robotic Mechanism with Integrated EPAM Actuators. Ebraheem I. Fontaine

Transcription

1 A Laboratory Demonstration of a Parallel Robotic Mechanism with Integrated EPAM Actuators by Ebraheem I. Fontaine Submitted to the Department of Mechanical Engineering in Partial Fulfillment of the Requirements for the Degree of Bachelor of Science in Mechanical Engineering at the Massachusetts Institute of Technology June Massachusetts Institute of Technology All Rights Reserved Signature of Author Department of Mechanical Engineering May 10, 2002 Certified by Steven Dubowsky Professor of Mechanical Engineering Thesis Supervisor Accepted by... Ernest Cravalho Chairman, Undergraduate Thesis Committee 1

2 A Laboratory Demonstration of a Parallel Robotic Mechanism with Integrated EPAM Actuators by Ebraheem I. Fontaine Submitted to the Department of Mechanical Engineering on May 10, 2002 in Partial Fulfillment of the Requirements for the Degree of Bachelor of Science in Mechanical Engineering ABSTRACT Planetary exploration will become a major focus of space exploration within the next several decades. In order to meet NASA s future goals, robots must be capable of performing complex tasks such as mapping terrain, constructing facilities, and collecting samples. Current planetary rovers are not capable of meeting these requirements, because their complex architectures and conventional components limit their capabilities. A new design concept for planetary robots has been proposed for highly reconfigurable robots that can perform a wide range of tasks. This concept of self-transforming explorers (STX) does not use conventional components such as motors, bearings, and gears. Instead, compliant mechanisms are embedded with binary actuation. In this thesis, a potential design for a STX using electrostrictive polymer artificial muscles as actuators is experimentally implemented. A two-stage robotic platform with six degrees of freedom and sixty-four discrete states is experimentally demonstrated. Thesis Supervisor: Steven Dubowsky Title: Professor of Mechanical Engineering 2

3 Acknowledgements I would like to thank everyone in the MIT Field and Space Robotics Laboratory for the wonderful experience and welcoming environment they provided for me. I mention a few that especially influenced my work. I thank Professor Dubowsky for the opportunity to work in his lab and conduct very interesting research that provided a strong addition to my undergraduate education; Moustapha Hafez, although no longer at MIT, for providing me with strong supervision and direction; Matt Lichter, who brought me into the FSRL as a UROP student, introduced me to the project, and always provided a open-ear; and Andreas Wingert for his recent supervision of my research, which certainly would not have been successful without him. I thank my family, Umme, Daddy, Muneera, Hajure, and Nargis, for their continued mental and financial support. This document represents the culmination of my undergraduate education. Finally, all praise is due to Allah, the Beneficient, the Merciful, for without him, nothing is possible. 3

4 Table of Contents ABSTRACT...2 Acknowledgements Introduction Self-Transforming Planetary Explorers (STX) Embedded Actuators Previous Research Binary Robotics Electrostrictive Polymer Artificial Muscles (EPAM) Summary Development of EPAM Actuator Fundamental Operating Principle Elastomeric Film Preparation Flexible Frame Actuator Characteristics Linear Bistable Element Summary Design of a Binary Robotic Device Kinematics and Functional Requirements Actuator Orientation Solid Model Mass and Displacement Predictions Results Conclusions and Future Work References Appendix A - Design of a High Voltage Mechanical Switch Appendix B - Experimental Setup

5 Table of Figures Figure 1.1 Simulation of Sojourner, Showing Difficulty Negotiating Terrain (Courtesy JPL)... 8 Figure 1.2 The Advantages of Self-Transforming Explorers (Andrews 2000)... 9 Figure 1.3 BRAID I Experimental Prototype (Oropeza 1999) Figure 1.4 BRAID II Experimental Prototype (Courtesy M. Lichter) Figure 1.5 EPAM Attached to a Rigid Frame (a) 0kV; (b) 6.5 kv (Courtesy A. Wingert) Figure 2.1 Illustration of EPAM Concept (Weiss 29) Figure 2.2 Stiffness Plot of Flexible Frame Figure 2.3 Partially assembled EPAM actuator (a) exploded view, (b) actual (Courtesy A. Wingert) Figure 2.4 Force-Displacement Plot of Partially Assembled EPAM Actuator Figure Linear Bi-stable Element (Courtesy A. Wingert) Figure 2.6 Force-Displacement Profile of Linear Bi-stable Element Figure 2.7 Creep of Linear Bi-stable Element Figure 2.8 Exploded View of Fully Assembled EPAM Actuator Figure 2.9 Force-Displacement profile of fully assembled actuator Figure 3.1 Illustration of Binary Robotic Concept (Courtesy M. Lichter) Figure 3.2 Concept Drawing of EPAM Actuator Orientation(Courtesy A. Wingert) Figure 3.3 Exploded View of One Stage BRAID III Design Figure 3.4 Illustration of Hinge Degrees of Freedom Figure 3.5 Illustration of BRAID III in two configurations Figure 3.6 (a) One Stage BRAID III with two actuators on; (b) Two Stage BRAID III with one actuator on in each stage (Courtesy A. Wingert) Figure 3.7 Schematic of Electrical Circuit Figure 3.8 Displacement and Current Profile of Single EPAM Actuator Figure 3.9 Force Predictions for Multi-Layered EPAM Actuator Figure A.1 High Voltage Switch; (a) Concept (Lichter 2001); (b) Design Figure B.1 - Schematic of Experimental Setup for Force-Displacement Measurements..48 5

6 Chapter 1 Introduction The goal of this thesis is to design, fabricate, and test an experimental system to demonstrate a new class of lightweight robotic devices that use non-conventional actuators. This work is part of an ongoing project at the MIT Field and Space Robotics Laboratory to develop concepts for self-transforming robotic planetary explorers (Andrews 2000; Lichter et al. 2000). Space exploration is an important area of current scientific research because knowledge about the universe will provide valuable information to meet challenges of the upcoming century. Over the next several decades, NASA plans to undertake increasingly complex missions for planetary exploration. These missions will require autonomous robotic systems to perform increasingly complex tasks, such as negotiating rough terrain, performing land surveys, collecting samples, and preparing resources for future human inhabitants. Current rover technology will be unable to meet these requirements. Current robots utilize conventional components such as gears, bearings, and fasteners and conventional methods of actuation such as electric motors and hydraulics. These components are typically made of metal, resulting in a heavy system that does not offer the versatility required for future space missions. A new paradigm in space robotics is needed that creates lightweight, inexpensive, robust, and 6

7 easy to control devices (Lichter 2001). This can be achieved by using compliant mechanisms and binary actuators made from smart materials. These robots will have reconfigurable features to allow them to adjust to various environmental obstacles they might face (Andrews 2000; Lichter et al. 2000). The funding for this project was provided by the NASA Institute for Advanced Concepts (NIAC) whose goal is to develop advanced concepts which will result in changes to the nation's future aerospace policies and plans (NIAC 2002). Although robust designs for all-plastic, lightweight, binary robots have many potential applications, their potential to expand the capabilities of planetary explorers is the focus of this thesis. Significant research has been done to develop previous experimental prototypes (See Section 1.3.1). These systems have identified many of the key challenges of designing a robot with compliant mechanisms and binary actuators. This thesis applied what was learned previously and redesigned a new system with improved performance. 1.1 Self-Transforming Planetary Explorers (STX) Planetary rovers such as the Sojourner have a very complex architecture made from conventional material and actuators that limit it to performing simple tasks (Lichter 2001). Figure 1.1 demonstrates some of the limitations of this rover design as it shows difficulty traversing the rock. 7

8 Figure 1.1 Simulation of Sojourner, Showing Difficulty Negotiating Terrain (Courtesy JPL) The concept of a self-transforming planetary explorer was developed at the MIT Field and Space Robotics Laboratory (Andrews 2000; Lichter et al. 2000; Lichter 2001). Unlike conventional robotic explorers, which are limited to a fixed configuration, these robots would be capable of transforming themselves into a wide variety of configurations. This increased adaptability would allow them to perform various tasks. Figure 1.2 illustrates some of the advantages of STX over current rover technology. One approach to developing self-transforming robots is to create binary structures that use flexible components, embedded actuators, and bi-stable mechanisms. With many binary actuators and many degrees of freedom, a large number of discrete states are achieved that approaches the behavior of continuous systems (Lichter 2001). Not only will binary devices create simple mechanical designs, they will also allow simple, robust control 8

9 systems. Because each actuator is either in an on or an off state, no feedback is needed to determine position. The open loop control eliminates the need for sensors. Figure 1.2 The Advantages of Self-Transforming Explorers (Andrews 2000) The potential success for binary robotic systems has been realized for some time (Chirikjian 1994). However, previous binary systems have used conventional components and were unsuccessful in creating effective actual physical systems. They were too heavy and resulted in impractical designs for the applications discussed in this thesis (Hafez 2002). 1.2 Embedded Actuators One of the fundamental challenges for developing robust binary systems is finding an appropriate actuator. Various non-conventional materials such as piezoelectric, electromagnetic, shape memory alloy, and electrostrictive polymers offer 9

10 alternative methods of actuation for binary systems. One actuator that offers strong potential for binary devices is called the Electrostrictive Polymer Artificial Muscle (EPAM) (Kornbluh 1999). They are given this name because some of their behavior is similar to that of biological muscle. They are very simple in construction, consisting of an elastomeric polymer sandwiched between two compliant electrodes. When a voltage is placed across the electrodes, they squeeze together and force the polymer to displace in a planar manner. Because of their simple construction, EPAMs are inherently lightweight and inexpensive. They can also be integrated into mechanical designs using few components. 1.3 Previous Research Binary Robotics Previous work in the MIT Field and Space Robotics Laboratory has demonstrated experimental prototypes for self-transforming planetary explorers using compliant mechanisms and smart material actuators. The first design of a Binary Robotic Articulated Intelligent Device (BRAID) is shown in Figure 1.3. This robotic manipulator is fabricated from polyethelyne and consists of five parallel stages that are serially connected by flexible linkages. Shape Memory Alloys (SMA) were used to actuate each flexible linkage joint, and search techniques for the forward and inverse kinematics of this device were evaluated (Sujan et al 2000). Another proposed component of this articulated device is bi-stable mechanisms. In order to remain in an extended position, the BRAID requires an actuation force. This would require continuous power supply to the actuators and unwanted energy 10

11 consumption. To keep the device in a fixed state, bi-stable or locking mechanisms can be employed to allow power requirements to be reduced. Although they were not incorporated into this design, their potential benefit was recognized. Figure 1.3 BRAID I Experimental Prototype (Oropeza 1999) A second generation BRAID was developed to explore new designs in compliant mechanisms, bi-stable mechanisms, and methods of actuation. It was fabricated from Delrin instead of polyethelyne for improved fatigue properties and machineability. The flexible linkages used cross-flexure hinges to achieve the necessary degrees of freedom. A fatigue analysis was performed to ensure that these members could endure 10 7 cycles (Hafez et al. 2002). Bi-stable mechanisms consisting of a thin buckled beam undergoing snap-through motion were incorporated. Electromagnetic actuation was also incorporated instead of SMA s. These actuators consisted of a pair of magnets with inverted polarities that were forced into two states by the induced magnetic fields of a 11

12 copper coil. These actuators produced forces of 1.4 N when a peak current of 3 Amps was used (Hafez et al 2002). Figure 1.4 BRAID II Experimental Prototype (Courtesy M. Lichter) Both the BRAID I and BRAID II incorporated embedded actuators into their final design. However, each method of actuation had properties that were obstacles to the practical implementation. The SMAs exhibited slow response time and small strains. The electromagnetic actuators exhibited a fast response time and adequate range of motion, however, had a poor power to weight ratio that limited the number of stages to two Electrostrictive Polymer Artificial Muscles (EPAM) Within the last decade, an increasing amount of research has been done in the area of electrostrictive polymers. The major thrust has been to transition EPAM actuators from a novel concept to practical applications. Significant work has been performed at SRI International to develop a theoretical model to predict the electrical and mechanical behavior of EPAMs (Pelrine et al 2000; Kofod et al 1999). Many ideas and devices for 12

13 EPAMs such as manipulators, couplings, rotary motors, biomimetic devices, inchworm robots, diaphragm pumps, and linear actuators have been reported (Kornbluh et al 1999; Cho et al. 2001; Eckerle et al 2001). Their properties, including high strains and high force to weight ratios, make them a promising actuator technology for use in high degree of freedom binary robotic systems (Bar-Cohen 2001). Figure 1.5 EPAM Attached to a Rigid Frame (a) 0kV; (b) 6.5 kv (Courtesy A. Wingert) Other research has attempted to characterize the similarity that exists between biological muscles and EPAMs and discuss the validity of this comparison in greater detail. Although EPAM s clearly possess muscle-like properties, there is a need for a standard means of comparison (Full 2000). There are certain behaviors that are not identical between EPAMs and biological muscles. Biological muscles contract when stimulated, while EPAMs increase in length when charged. In order to achieve musclelike contractions, one actuator must be charged continuously to keep it in a relaxed state, then discharged to make it contract and exert a force on the environment (Meijer et al 2001). The term electrostrictive is used to describe the mechanical behavior of a material in an electric field. Ronald Pelrine and others feel that electrostrictive is a more appropriate description than electrostatic for these actuators because the dielectric and 13

14 mechanical properties of the polymer material determine the magnitude of the stress and strain response (2000). 1.4 Summary A Binary Robotic Articulated Intelligent Device (BRAID) that incorporates EPAM actuators is a novel device that can potentially meet the requirements of NASA planetary exploration in the next several decades (Wingert 2002). It will be lightweight, deployable, and able to perform complex tasks that are currently not feasible for planetary rovers. This thesis will discuss the design and fabrication methods for a selfcontained EPAM actuator and assess its fundamental characteristics through experimental testing. A design for a third generation BRAID that incorporates EPAM actuators is discussed. An experimental prototype of a two stage robot with six degrees of freedom is demonstrated and tested for performance. 14

15 Chapter 2 Development of EPAM Actuator 2.1 Fundamental Operating Principle Electrostrictive Polymer Artificial Muscles (EPAM) operate using a simple principle. Muscle-like actuation occurs when a potential difference is placed across two compliant electrodes that have an elastomeric polymer between them. The electric field causes the oppositely charged electrodes to attract each other and decrease the thickness of the film. It also causes the identically charged electrodes to repel each other and force the elastomeric film in a planar direction (See Figure 2.1). This coupling effect creates an effective pressure on the film known as the Maxwell pressure due to the compliance of both materials. The governing equations of the EPAMs have been reported by Pelrine, Kornbluh, and Joseph at SRI International (1998). Below is a brief summary of their conclusions. The EPAM acts as a capacitor as charge is stored on the electrodes, Figure 2.1 Illustration of EPAM Concept (Weiss 29) 15

16 which are separated by the film. The capacitance, C, is equal to ε A C o ε = t (1) whereε o is the permittivity of free space, ε is the dielectric constant, A is the surface area of the electrodes, and t is the thickness of the film. Using Equation 1, the resulting effective pressure, p, can be derived as 2 2 ε oεv p = ε oεe = (2) 2 t where E is the electric field and V is the applied voltage. Assuming a constant volume is maintained during expansion, a relationship for the planar strains, s x and s y and the thickness strain s z can be written as ( 1 )( 1+ s )( 1+ s ) = 1 + x y z s. (3) Electrostrictive actuators can be made from a variety of polymers and electrodes. Some of these materials include polyurethane, silicone, fluorosilicone, and acrylic as the elastomeric film and carbon, silver, and gold as the compliant electrode. The ideal polymer will have a high breakdown electric field, a high dielectric constant, and a low modulus of elasticity. The breakdown electric field and strain of the EPAM can be increased by pre-stretching the material (Kofod et al. 2001; Kornbluh et al. 2001; Pelrine et al 2000). However, stretching the material in a planar direction decreases its thickness, which results in a smaller dielectric constant. It also increases the modulus of elasticity, which causes a decrease in strain for a given pressure applied by the electrodes. The breakdown electric field is the most significant property of the elastomeric film because once this value is reached, the actuator is effectively destroyed. The internal arcing causes a crack in the material which propagates and causes the entire film to rip (Kofod et 16

17 al. 2001). Although these properties cannot be maximized simultaneously, a non-uniform pre-stretch can cause significant strain in one of the in-plane directions. If the material is stretched greater in one direction and less in the other, it will tend to deform in the less strained direction when charged. Tests have shown that 3M s VHB 4910 acrylic has the highest stress and strain performance out of all elastomeric films tested to this date. When pre-stretched 75% in the active direction and 540% in the inactive direction, it exhibited linear strains of up to 215% and effective pressures of 2.4 MPa (Pelrine et al, 2000). 2.2 Elastomeric Film Preparation Although various types of polymers and electrodes can be used to construct EPAMs, this design utilizes two layers of acrylic tape as the polymer and three layers of silver grease as the compliant electrode. The proper assembly of these actuators is fundamental to their high performance. Prior to adhesion with the flexible frame, the acrylic film must be stretched from its original stock size. The film is stretched greater in the passive direction and less in the active direction of the actuator. The active and passive directions of the actuator are denoted in Figure 2.2 by the y and x axes respectively. Stretching the film greater in the passive direction causes an increased stiffness, which forces most of the material to displace in the active direction when it is actuated. The original stock is packaged in a 36 yard long roll of tape. The width and thickness of the acrylic film are 25.4 mm and 1.02 mm respectively. In order to stretch the acrylic to the proper size to fill the dimensions of the frame, an initial segment 30 mm in length is used. It is then stretched 433% in the passive direction to a length of 160 mm 17

18 and 226% in the active direction to a length of 35 mm. The final thickness of the film is approximately 0.06mm, which can be approximated using Equation 3. The acrylic is stretched using a manual sliding vice. 2.3 Flexible Frame The flexible frame provides another essential element to the design of the EPAM actuator. It provides the appropriate boundary conditions on the film and protects it from tearing. The stiffness of the frame was chosen to provide an appropriate restoring force on the stretched film. Exposed edges of the film decrease the performance of the EPAM because it results in a non-uniform thickness. The larger thickness in the center causes a decrease in the local electric field, which causes lower strains and forces (Kofod et al. 2001). At high enough voltages, arcing can also occur around exposed edges and cause the actuator to malfunction (Weiss 39). The enclosed flexible frame eliminates both of these problems. Delrin was chosen as the material for this application because it has excellent fatigue properties and can be fabricated easily. There are several important considerations when designing the frame. One is that an adequate surface area must exist around the border of the frame for the film to adhere to. Otherwise, the film will pull through and the actuator will fail. An additional factor is the minimum hinge thickness, t that occurs within the flexible frame. Because the stiffness of cantilevered beam is proportional to t 3, it can dramatically change with small variations in the value of t. The flexible frames were fabricated using abrasive water-jet machining because of the fast machining time, which enabled several design iterations in a short period of time. 18

19 However, the limitations of the abrasive water-jet cutter prevented manufacturing Delrin flexural hinges with a thickness of less than 0.5mm. Flexible frames with a hinge thickness of 0.5mm were manufactured consistently and, thus, actuators with identical stiffness values and performance were produced. The linear stiffness range of the flexible frame exists when the frame is in a compressed state. When the frame is expanded beyond its neutral position, the stiffness increases exponentially as shown in Figure 2.2. Figure 2.2 Stiffness Plot of Flexible Frame According to Figure 2.2, the frame exhibits nearly perfect elastic behavior in its region of compression. However, in the region of strain beyond y o, there is visible energy loss due to hysteresis because the frame does not return to the same initial force value. In order to utilize the flexible frame in its linear stiffness region, the frame is compressed approximately 23 mm to ensure the actuator operates within this desired region. Once compressed to the desired value, the frame is attached to the pre-stretched acrylic film using silicone adhesive. Previous EPAM actuators used super-glue adhesive; however, this did not form a durable bond between the acrylic and the Delrin. As a result, the 19

20 actuators often failed within a short time period ranging from several cycles to a few days. The silicon adhesive is also flexible and compliant itself, which allows it to properly seal the entire frame and reduce stress concentrations in the acrylic. An actuator using two layers of acrylic film is used in this design because it provides a greater force than a single layer. Compliant electrode is brushed onto both sides of the acrylic films using a soft applicator, and thin copper tape is used to create electrical leads to both the inner and outer electrodes. The high voltage is applied to the inner electrode, and the two outer electrodes are grounded. The partially assembled EPAM actuator is shown in Figure 2.3. (a) 20

21 (b) Figure 2.3 Partially assembled EPAM actuator (a) exploded view, (b) actual (Courtesy A. Wingert) 2.4 Actuator Characteristics The force-displacement characteristics of the actuator are plotted in Figure 2.4. The plot contains three separate curves. The first and second curves show the actuator s force displacement profile for 0 kv and 5.5 kv respectively. There is a small amount of hysteresis visible in each plot as the arrows indicate that the actuator does not follow the same profile during expansion and contraction. Experiments show that the force generated by the EPAM during expansion depends on the velocity (Full et al. 2000). When tested using the experimental setup shown in Appendix B, the EPAM was expanded at a slow rate to minimize losses due to viscoelastic properties of the film and flexible frame. This energy loss is equal to the area between the separated lines. The stroke of the actuator is approximately 5.5mm, which equates to a strain of 31%, and the static force achieved by the actuator is approximately 2 N. 21

22 Figure 2.4 Force-Displacement Plot of Partially Assembled EPAM Actuator The third curve in Figure 2.4 represents the actuator workcycle. The plot is obtained by turning the voltage (5.5kV) on and displacing the actuator until the value of the force is zero, then turning the voltage off and returning the actuator to its initial state where the force also reads zero. During both the charged and uncharged states the actuator stiffness remains a constant value of 0.57 N/mm. The work done by the actuator through one cycle is equal to the area enclosed by the work cycle, which is approximately 11 mj. This elastic behavior exhibited by the actuator is not ideal for many applications. Typical actuators, such as motors or hydraulic pistons, have constant force output 22

23 throughout their displacement. In the case of the actuated EPAM, the force linearly decreases to zero as it moves throughout its stroke. The next section discusses the solution implemented to address this concern. 2.5 Linear Bistable Element The final component of the EPAM actuator is the linear bi-stable element (LBE). This element is attached in parallel with the EPAM actuator and provides a negative spring coefficient to offset the stiffness of the acrylic film and flexible frame and provide a constant force region in the actuator stroke. It also improves actuator performance by increasing the displacement of the actuator by nearly 4mm. The design of the LBE consists of two parts: a base with two flexure arm components and an oversized insert, which forces the flexure arms into two stable states. A picture of these components displaying the two stable states is shown in Figure 2.5. Figure Linear Bi-stable Element (Courtesy A. Wingert) The two parts are mated using a dovetail design, which prevents the insert from slipping out during the stroke, while still allowing easy removal. The stiffness of the 23

24 linear bi-stable element can be modified by making slight changes to the geometry of the components. The insert is designed with a larger width than the distance between the two flexure arms and this offset distance is what forces the LBE into two stable states. By increasing the width of the insert, the stiffness and the linear stiffness range of the LBE is increased. Typical offset distances ranged from 2.5mm to 4.5mm. Another dimension that is modified to change the stiffness properties of the LBE is the beam thickness of the base. Decreasing the beam thickness of the base also decreases the stiffness of the LBE. In order to optimize the performance of the EPAM actuator, a LBE with a stiffness value of approximately N/mm and a large linear range is desired. This negative stiffness coefficient will cancel the positive stiffness of the flexible frame to produce a constant force throughout the stroke. The LBEs were fabricated from a single piece of Delrin using the abrasive water jet cutter. This rapid fabrication method allowed for experimental stiffness values of the LBE to be determined with little turn around time. LBEs with various offset distances and beam thicknesses were measured experimentally until the desired stiffness value was achieved. Figure 2.6 shows the plot of the force displacement curve of the final LBE design used in the EPAM actuator. The LBE displays a nominal amount of hysteresis throughout its stroke, which is due to energy loss in the flexure hinges. 24

25 Figure 2.6 Force-Displacement Profile of Linear Bi-stable Element The stiffness value of N/mm was the closest value to 0.57 N/mm achieved with the design, and the linear range was approximately 6 mm. The location where the curve crosses the neutral axis (zero force) represents the expanded stable state of the LBE. The compressed stable state is reached due to the beam providing a mechanical stop for the flexure arms. The region between these states, where it has a negative stiffness constant, is what the design exploits. One shortcoming of using plastic flexure components is that they are susceptible to creep due to material degradation. Because the LBE relies on deformation of the flexure arms to reach its stable state, the Delrin will undergo creep if it is left in the deformed state for a significant amount of time. The graph below in Figure 2.8 displays the effect of leaving the LBE in a stable state. 25

26 Figure 2.7 Creep of Linear Bi-stable Element The stiffness of the LBE was measured on three different occasions and each exhibited a decrease in stiffness from the previous measured value. The original stiffness of the LBE was measured as N/mm. This value decreased to N/mm and N/mm after 19 hours and 44 hours, respectively, when it was kept in its stable state. This presents a problem to our actuator design because it requires the LBE to have the same magnitude of stiffness as the acrylic film and flexible frame, approximately 0.57 N/mm. Because this behavior is a fundamental consequence of the material and design used, the simplest solution is to ensure that the bi-stable insert can be easily removed when the actuator is not in use. 26

27 The fully-assembled actuator incorporates both the EPAM embedded in the flexible frame and the linear bi-stable element. Figure 2.8 shows an exploded view of the various components. Figure 2.8 Exploded View of Fully Assembled EPAM Actuator The force-displacement of the actuator is shown below in Figure 2.9. The graph verifies the predicted behavior of the actuator, which is a graphical summation of the values shown in Figures 2.4 and 2.6. When 5.5 kv is applied, the actuator exhibits a constant force difference of 1.5 N through a stroke of 15 mm to 22 mm. From 22 mm until the end of its stroke, the actuator exhibits the behavior of the flexible frame as the force constantly decreases to zero. This occurs because the actuator has moved outside of the linear stiffness range of the LBE. 27

28 Figure 2.9 Force-Displacement profile of fully assembled actuator As the arrows on the 0 kv and 5.5 kv plots indicate, there is a significant amount of hysteresis present in the fully-assembled actuator. It is a summation of all of the energy losses of each component. The plot of the work cycle indicates that the actuator has greater force in pushing than in pulling. When the high voltage is applied, the average force applied during the expansion is 1.5 N, however, when the high voltage is removed, the average force applied during contraction is only 0.5 N. 28

29 2.6 Summary In this section, actuator development that utilizes electrostrictive polymers was described. Andreas Wingert of the MIT Field and Space Robotics Laboratory developed this EPAM actuator design that is integrated into the BRAID III device discussed in the next chapter (Wingert 2002). These actuators use pre-stretched acrylic film embedded in flexible frames and linear bi-stable elements to achieve linear strains of 47% or 9 mm and forces of 1.5 N. The total weight of the actuator is 9.2 grams, and the weight of the active area is a mere 1.2 grams. With such small mass properties, these actuators have the ability to lift objects several times their weight. Given their characteristics, it makes them an ideal actuator for lightweight, binary robotic applications. 29

30 Chapter 3 Design of a Binary Robotic Device 3.1 Kinematics and Functional Requirements The development of the EPAM actuators discussed in the previous section will motivate the design of the robotic device described in this section. The design of the EPAM actuated BRAID III is based on previous designs developed at the MIT Field and Space Robotics Laboratory (Hafez 2002, Sujan 2001, Lichter 2001, Oropeza 1999). A single stage parallel platform device, as shown in Figure 3.1, will have 2 3 or 8 different stable states based on the binary nature of the linkages. Figure 3.1 Illustration of Binary Robotic Concept (Courtesy M. Lichter) A device that operates on the same principle, but uses EPAM actuators will have similar kinematics The forward kinematics of this structure have been determined by Matt 30

31 Lichter (2001). The configuration requires that the upper joint have two degrees of freedom and the lower joint have one degree of freedom. In order to meet the design criteria of developing a lightweight, binary robotic device with EPAM actuators, several functional requirements must be taken into account. First, the actuators must be packaged properly to minimize the overall volume. Second, the weight of each component must be minimized to reduce the overall robot weight. Third, the components must have a modular design that allows for ease of assembly and ease of replacement should a component fail. 3.2 Actuator Orientation In designing the EPAM actuator orientation of the with respect to the parallel platforms, two initial concepts arose that are illustrated in Figure 3.2. Although both concepts will work functionally, concept (B) occupies 67% less volume. The flexible frames increase to approximately 90mm in width when compressed and attached to the LBE. This value is used to calculate the cross-sectional area of the BRAID to give an indication of the volume, since each concept would have virtually the same height. Having a minimum volume occupied is important for space applications. Concept (B) was also more desirable because of the open space inside of the actuators. This space was needed because we anticipated placing the high voltage wires to charge the actuators through the center of the structure. 31

32 Figure 3.2 Concept Drawing of EPAM Actuator Orientation(Courtesy A. Wingert) 3.3 Solid Model The kinematics of the design illustrated in Concept (B) requires two different joints. The upper joint, where the top of the actuator and the upper platform are joined, requires a joint that allows motion about two axes, and the lower joint, where the bottom of the actuator and the lower platform are joined, requires a revolute joint that allows motion about one axis. The following design, shown in Figure 3.3, was implemented in order to meet all of the functional requirements and kinematics listed above. The design is motivated by concept (B) and is made entirely from Delrin using the abrasive waterjet cutter. Each EPAM actuator is located symmetrically about the center of the platform at 120 degree intervals. This configuration allows for multiple stages of the BRAID III to be stacked on top of each other and offset by 60 degrees. The platform has two different 32

33 mating areas that are used for either the upper or lower hinge assembly. The upper hinge has a Figure 3.3 Exploded View of One Stage BRAID III Design dovetail shape that mates with the platform using press-fit, and the lower hinge assembly mates to a rectangular shape in the platform also using press-fit. The lower hinge assembly creates a one degree of freedom rotational hinge by mating two opposite beam flexures to create a cross-flexure hinge. It also contains an area that functions as a rectangular insert used to position the EPAM and LBE perpendicular to the platform. The upper hinge assembly is a two degree of freedom hinge that approximates a spherical hinge for the small angles of deflections that the BRAID III operates in. It contains four parts machined from Delrin, which are glued to a piece of plastic film. The dovetail shape at one end mates with the platform and the other side is press-fit into a slot in the flexible frame. As illustrated in Figure 3.4, both hinges achieve rotational freedom about the X-axis and the upper hinge also achieves rotational freedom about the Y-axis. 33

34 Figure 3.4 Illustration of Hinge Degrees of Freedom The LBEs were placed on the outside of the structure to allow ease of access in order to remove the insert when not in use and prevent mechanical creep of the flexure arms. The outer flexible frame was fashioned with a tapped hole to attach a M3 plastic fastener. This fastener rigidly attaches the bi-stable insert to the flexible frame and allows for proper adjustment before use. The overall design was successful in creating a modular design with interchangeable components. All components assemble in a logical manner to prevent incorrect assembly and malfunctioning. Should a component fail, all parts are mated by press-fit to allow for easy removal and replacement. 34

35 3.4 Mass and Displacement Predictions Using Solid Works solid modeling software, the total volume of the structure was estimated as 28.1 cm 3. An initial estimate of the mass of a single stage is 38.8 grams, given that Delrin has a density of about 1.41 g/cm 3, and the acrylic has a density of 0.96 g/cm 3. Given that the actuators can achieve a displacement of 9mm, the angular tilt that the upper platform can achieve are calculated by assuming the actuator moves through the arc of a circle with radius, R, which is the distance between the axis of rotation and the actuator. The following table contains the initial estimates of the angular tilt, θ that will be achieved. Number of Actuators R (mm) θ (degrees) turned on ; Y=9mm Table 1 Estimates of Platform Displacement The number of stages is limited by the actuator capability. The lower stages have to support the weight and moment of the upper stages. Given a static force of 1.5 N that a two layer actuator can produce, an initial estimate of the number of stages that the BRAID III can support was made. Each stage has a mass of 36.3 grams. Using a force balance approximation, it was found that a two stage BRAID was feasible with the current actuators. Adding additional layers of film increases actuation force without adding significant mass allowing for more stages. Figure 3.5 illustrates a predicted model of two configuration of the BRAID III given a displacement of 9mm for one actuator in each stage. 35

36 The project goal was to have two working stages of the design to demonstrate six degrees of freedom and 2^6=64 discrete states. Figure 3.5 Illustration of BRAID III in two configurations 3.5 Results A successful prototype of the BRAID III design containing two functional stages was developed. The prototype weighed 36.3 grams per stage, which is within 2% of the predicted mass by the solid model. A fully assembled BRAID III with one and two stages is shown below in Figure 3.6. The electrical connections to the actuators were made by soldering a wire to the copper tape and directing all of the electrical cables through the center of the BRAID III. All of the ground wires were tied together to create a common ground, and the high voltage wires were charged with 5.5 kv independently. 36

37 Because the BRAID III operates at such a high voltage, electrical shielding is important because arcing can damage the electrostrictive polymer and cause it to malfunction. The thin insulation on the wire was sufficient protection against arcing, however, the exposed ends had to be placed at least a couple of centimeters apart to prevent arcing. Figure 3.6 (a) One Stage BRAID III with two actuators on; (b) Two Stage BRAID III with one actuator on in each stage (Courtesy A. Wingert) In general, the electrical control of a binary robotic device using EPAM actuators is difficult to achieve because current semiconductor devices are not optimized to handle their requirements. They present high impedance loads and are highly capacitive. Not only do they require high voltages but also high peak currents for fast actuation (Eckerle 2001). Because electrical control was not the goal of this research, and the robot was binary in nature meaning that each actuator was turned either on or off, a mechanical switch was fabricated to switch the actuators on and off. Because only two high voltage power supplies were available, this high voltage switch had to be capable of handling 5.5 kv without arcing. One high voltage power supply was used for each stage. The details 37

38 of this switch design are presented in Appendix A. The overall circuit design that was implemented for two stages of the BRAID III is shown below in Figure 3.7. Figure 3.7 Schematic of Electrical Circuit Each of the six actuators is modeled as a capacitor whose value can be calculated using Equation 1. The value of R 1, the resistor placed in series with the EPAM, is 10 MΩ and the value of the resistor R 2, where all of the actuators are grounded to is 50MΩ. When the actuators are turned off and begin to discharge, these resistors increase the RC time constant of the circuit and prevent damage to the EPAM from rapid discharging. Another impressive characteristic of the EPAM actuators is their low power consumption. Figure 3.9 plots the displacement and current of one actuator as a function of time. At about 2.5 seconds, the actuator is charged and the lower graph shows a peak current of 32 µa drawn by the actuator. The current then rapidly decreases to a steady state value of 2.7 µa. Since the actuator is charged using 5.5 kv, the peak and steady state power consumptions of each actuator are calculated as 0.18 W and W respectively. The upper profile shows a rapid increase in displacement when the voltage 38

39 is initially applied. The actuator reaches about 95% of its final value within 3 seconds. This response time is largely determined by the electrical circuit used to charge and discharge the actuator. Others have shown response times on the order of 10 ms for EPAM actuators, and predict a decrease to less than 1 ms with the optimization of an electrical control circuit (Full et al. 2000). 24 actuator displacement and current 22 length-y (mm) time (s) current (ma) time (s) Figure 3.8 Displacement and Current Profile of Single EPAM Actuator An unattractive characteristic of these actuators is also illustrated in the upper profile. From time 6 to 11 seconds, the actuator continues to increase in length even though the current has reached a steady state value. In this region, the force is also decreasing to zero as shown in Figure 2.9. This creeping effect is very slight, with less than a millimeter in displacement within 5 seconds. This can be addressed by limiting the actuator stroke with a mechanical stop. This will provide repeatable endpoints and, thus binary action. When the voltage is turned off at 11 seconds, there is a negative 39

40 current surge, which is caused by the actuator discharging. The rate at which the actuator discharges is again determined by the RC time constant of the circuit. Also, the rate of change in the displacement is less when the voltage is turned off, than when it is turned on. This is the result of what was shown earlier in Figure 2.9. The force applied by the actuator is less in contraction than in expansion due to the relative placement of the LBE and flexible frame. The final weight and performance specifications of the BRAID III are listed below in Table 2 and compared to a previous generation BRAID II device developed by Moustapha Hafez and Matt Lichter that utilizes electromagnetic actuation. Specifications BRAID III BRAID II Weight per Stage 38 g 73.6 g Weight of Single Actuator 1.2 g (acrylic film and electrode) 14.0 (magnets and coil) Max. Force 1.5 N 1.4 N (per actuator) Electrical Requirement 5.5 kv, 0.032mA 3.5 V, 3 Amps Bi-stable No Yes Table 2 Comparison Chart between BRAID III and BRAID II A multi-layered EPAM concept was demonstrated in BRAID III. Only two layers of film were used in the actuators, primarily due to manufacturing and assembly limitations. This proves to be an important area of future research because the force of these actuators can be dramatically increased by layering multiple films. In Figure 3.10, the predicted maximum force of the actuator as a function of its weight is shown for increasing layers. Based on this graph, forces up to 7.5 N can be achieved with an 40

41 actuator containing ten layers of film. An actuator of this size would require a flexible frame that is five times as stiff in order to support the pre-tensioned load of the acrylic. A LBE with increased stiffness would also be required to provide motion amplification and a constant force region throughout the stroke of the actuator. Adding layers and increasing the stiffness of the flexible frame would add additional mass to each actuator, but the force to mass ratio of 0.16 N/grams is expected to increase. Figure 3.9 Force Predictions for Multi-Layered EPAM Actuator 41

42 Chapter 4 Conclusions and Future Work A two stage binary robotic device with electrostrictive polymer artificial muscle actuators has been developed. This device demonstrated six degrees of freedom and 64 discrete states by manipulating a mirror payload of 25.3 grams. The entire structure weighs 73.0 grams and exhibited forces of 1.5 N per actuator. The device has a power consumption of W at steady state and a peak power of 0.18 W when the actuator is initially turned on. The actuators displayed a time response of about 3 seconds and exhibited some mechanical creep once the current reached steady-state. Significant improvement was made in the redesign of the BRAID. BRAID III had a mass nearly half that of BRAID II, which was mainly due to the alternative method of actuation. The EPAM actuators were 1/12 as massive as the electromagnets, and provided a 7% increase in the maximum force. The two stage BRAID III was also significantly stiffer than its electromagnetic counterpart. The press-fit hinges and flexible frames were more resistant to torsion deflections, while the cross flexure linkages and ball bearing spherical joints of BRAID II provided for a less rigid structure. Like the BRAID I, the experimental prototype demonstrated requires the continuous force of the actuator to keep it in an expanded state. Although EPAM actuators draw little power when charged continuously, an ideal design would incorporate 42

43 bi-stable mechanisms that will lock and hold each linkage in a fixed state, so that power can be removed from the system. A potential BRAID IV that makes improvements on the EPAM actuation and incorporates bi-stable mechanisms is needed to fully address all of the criteria identified by the self-transforming robotic planetary explorers project. Significant future work should be done in the area of EPAM actuators. These actuators provided an improved means of actuation for this application than the SMAs used in BRAID I. New ideas and techniques for manufacturing actuators will improve their performance and allow for multi-layering. Also, development in digital circuitry will permit proper electrical control of these actuators, which provide a challenge for conventional circuitry with their high impedance and high capacitive properties. In addition, new techniques for packaging EPAMs into linear actuators should be explored. A flexible frame is demonstrated in this design, but other geometries are possible. Since EPAM behave like a spring, new techniques for providing a constant force throughout its stroke should be explored. The linear bi-stable element is demonstrated in this discussion; however, other techniques are yet to be found. The suggestions listed above only highlight a few specific areas that should be evaluated in the short term to improve the BRAID design. These suggestions focus on the EPAM actuator because the actuator module represents the single most difficult obstacle in the creation of Self-Transforming Explorers. Yet there are much larger scale issues that must be addressed by new discoveries in order to advance the field. As these issues of design, manufacturing, materials science, and electrical control are addressed in the future, the idea of a Self-Transforming Explorer will transfer from concept to fullscale implementation. 43

44 References Andrews K. Elastic Elements with Embedded Actuation and Sensing for Use in Self- Transforming Robotic Planetary Explorers. M.S. Thesis, MIT, Cambridge, MA, Bar-Cohen, Y. Transition of EAP material from novelty to practical applications are we there yet? in Smart Structures and Materials 2001: Electroactive Polymer Actuators and Devices, Yoseph Bar-Cohen, Editor, Proceedings of SPIE Vol. 4329, p.2-8, March Chirikjian G S. A Binary Paradigm for Robotic Manipulators. International Conf. on Robotics and Automation, , Proc. IEEE Cho, Sunghwi, Sungmoo Ryew, Jae wook Jeon, Hunmo Kim, Jae-do Nam, Hyoukryeol Choi, and K.Tanie, "Electrostrictive Polymer Actuator Using Elastic Restoring Force," Proceedings of International Symposium on Robotic(ISR `2001), pp , Eckerle, J., S. Stanford, J. Marlow, J. Marlow, R. Schmidt, S. Oh, T. Low, S. Shastri A Biologically Inspired hexapedal Robot Using Field-Effect Electroactive Elastomer Artificial Muscles, in Smart Structures and Materials 2001: Electroactive Polymer Actuators and Devices, Yoseph Bar-Cohen, Editor, Proceedings of SPIE Vol. 4329, March Full R., K. Meijer, Artificial Muscles Versus Natural Actuators from Frogs to Flies in Smart Structures and Materials 2000: Electroactive Polymer Actuators and Devices, Yoseph Bar-Cohen, Editor, Proceedings of SPIE Vol. 3987, p.2-9, March Hafez, M., M. Lichter, S. Dubowsky. Optimized Binary Modular Reconfigurable Robotic Devices Proceedings of Proceedings of the 2002 IEEE International Conference on Robotics and Automation, Washington D.C. May Kofod, G., R. Kornbluh, R. Pelrine, P. Sommer-Larsen, Actuation response of polyacrylate dielectric elastomers, in Smart Structures and Materials 2001: Electroactive Polymer Actuators and Devices, Yoseph Bar-Cohen, Editor, Proceedings of SPIE Vol. 4329, p , March Kornbluh R., R. Pelrine, J. Joseph, R. Heydt, Q Pei, S. Chiba, High-field electrostriction of elastomeric polymer dielectics for actuation, in Smart Structures and Materials 1999: Electroactive Polymer Actuators and Devices, Yoseph Bar- Cohen, Editor, Proceedings of SPIE Vol. 3669, pp ,

45 Lichter, M.D., Sujan, V.A., Dubowsky, S., "Experimental Demonstrations of a New Design Paradigm in Space Robotics",Proceedings of the Seventh International Syposium on Experimental Robotics (ISER '00), December Lichter M. Concept Development for Lightweight Binary-Actuated Robotic Devices, with Application to Space Systems. Thesis for Masters of Science in Mechanical Engineering, Massachusetts Institute of Technology, May Meijer K., M. Rosenthal, R. Full, Muscle-like actuators? A comparison between three electroactive polymers, Smart Structrues and Materials 2001: Electroactive Polymer Actuators and Deevices, Yoseph Bar-Cohen, Editor, Proceedings of SPIE Vol. 4329, pp.7-15, March NASA Institute for Advanced Concepts (NIAC). The Institute. Online Posting, 25 April Oropeza G. The Design of Lightweight Deployable Structures for Space Applications. Thesis for the Bachelor of Science in Mechanical Engineering, Massachusetts Institute of Technology, May Pelrine, R., R. Konrbluh, J. Joseph, Electrostriction of polymer dielectrics with compliant electrodes as a means of actuations, Sensor and Actuators A: Physical 64, pp Pelrine, R., R. Kornbluh, Q. Pei, J. Joseph, High Speed Electrically Actuated Elastomers with Strain Greater Than 100%, Science, Vol. 287, pp , Sujan, V, M. Lichter, S. Dubowsky, Lightweight Hyper-redundant Binary Elements for Planetary Exploration Robots, Proc IEEE/ASME International Conference on Advanced Intelligent Mechatronics (AIM '01) 811, Como, Italy, July Weiss, P. Development and optimization of Electrostrictive Polymer Artificial Muscles actuators with large expansions and high forces. Thesis for Masters of Science in Mechanical Engineering, Ecole d Ingénieurs, Paris, July Wingert, A. Development of a Polymer-Actuated Binary Manipulator. Thesis for Masters of Science in Mechanical Engineering, Massachusetts Institute of Technology, May

46 Appendix A Design of a High Voltage Mechanical Switch Because only two high power supplies were available, a high voltage mechanical switch had to be constructed in order to charge the EPAM actuators. An ideal approach would have been to switch the actuators on and off at a low voltage and then amplify the signal to a high voltage. However, this would require six independent high voltage power supplies. Instead a simple mechanical toggle switch (also a bi-stable mechanism) was designed based on the principle illustrated below. An elastic element or tension spring is attached to the lever and secured at a point below the lever point of rotation. This causes the lever to become bi-stable. Figure A.1 High Voltage Switch; (a) Concept (Lichter 2001); (b) Design 46