Flexinol as Actuator for a Humanoid Finger - Possibilities and Challenges. Master thesis (60 pt) Øyvind Fjellang Sæther - PDF

Description
UNIVERSITY OF OSLO Department of Informatics Flexinol as Actuator for a Humanoid Finger - Possibilities and Challenges Master thesis (60 pt) Øyvind Fjellang Sæther Abstract Robots become more

Please download to get full document.

View again

of 101
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Information
Category:

Funny & Jokes

Publish on:

Views: 31 | Pages: 101

Extension: PDF | Download: 0

Share
Transcript
UNIVERSITY OF OSLO Department of Informatics Flexinol as Actuator for a Humanoid Finger - Possibilities and Challenges Master thesis (60 pt) Øyvind Fjellang Sæther Abstract Robots become more and more common in our every day lives as technology develops. Robots are normally actuated by pneumatics, hydraulics or servo motors. These technologies are mature and widely used, but other less commonly used actuators are also available. Among these we find the artificial muscle fiber Flexinol which belongs to a class of materials known as Shape Memory Alloys. This thesis aims to implement the artificial muscle fiber Flexinol as actuator for a humanoid finger. The first part of the thesis focuses on testing of single Flexinol wires to determine in what degree these are suitable for long term use as actuators. A test frame is built to investigate contraction speed, force and displacement for wires in different setups. Among these are tests with a small dead weight, a large dead weight, an antagonistic setup and a setup with a spring working as a passive antagonistic force. The second part of the thesis makes use of Flexinol as actuator when designing and prototyping a humanoid finger. The human finger is used as inspiration in this part, applying tendons and muscles in a human-like way. The finger is designed with CAD-software and then printed in plastic. It is then assembled with tendons and actuated with three Flexinol wires. Finally, an attempt to control the humanoid finger is done. Specially designed software and hardware is developed through the thesis to implement working experiments. Software for both a laboratory computer and a microcontroller is written to control the system and to collect sensory data respectively. I II Preface This thesis is part of my Master Degree at the University of Oslo, Department of Informatics. The thesis was carried out during 2008 in the research group Robotics and Intelligent Systems (ROBIN). First of all I want to thank my supervisor, Associate Professor Mats Høvin, for creative input, for motivating me to be creative and for valuable feedback both during my practical work and during writing. I also want to thank the following (sorted by topic): Vegard Friis Ruud, Charlotte Kristiansen, Marie Klemsdahl Eklund and Marte Lødemel Henriksen for fruitful discussions regarding the human anatomy, Kjetil Stiansen for help regarding practical and theoretical electronics and Andreas Gimmestad for his linguistic abilities. Thanks also go to friends and family for showing interest, specially to my father Dag Henning Sæther for guidance both during my practical work and during writing. Øyvind Fjellang Sæther November 2008 III IV Contents 1 Introduction Humanoid Hands Robot Hand Actuation Flexinol - Artificial Muscle Fibers Thesis Overview Short Conclusion Background Anatomy of The Human Hand Skeleton Tendons and Muscles Robotic Approach to the Human Hand Traditional Actuators Hydraulics Pneumatics Servo Motors Stepper Motors Electric Solenoids Intelligent Materials used as Actuators Shape Memory Alloys in General Flexinol Electroactive Polymers Actuator Comparison Power to Weight Ratio Feedback Sensors Displacement Transducers Force Transducers Used Tools Atmel AVR Microcontrollers I/O-Ports Memory Interrupts Counters and Pulse Width Modulation (PWM) Universal Synchronous and Asynchronous Serial Receiver and Transmitter (USART) Analog to Digital Converter (ADC) Watchdog Timer Clock Source Keithley KUSB Microsoft Robotics Studio Overview Concurrency and Coordination Runtime (CCR) Decentralized Software Services (DSS) Visual Programming Language (VPL) V 4 Own Methods Testing of Flexinol Fixation Test Degeneration Test Flexinol Antagonist Spring Antagonist PWM-controlled Test Frame Electronics Design Calibration Test Software Software for the Test Frame Web Application for Remote Surveillance Humanoid Finger Design Anatomical Model D Design Humanoid Finger Application Mechanical Design Electrical Schematics Communication Microcontroller Program Computer Interface Program Interface for Microsoft Robotics Studio Summary of Own Methods Experiments Calibration Results Displacement Calibration Force Calibration Testing of Flexinol Test Software Fixation Test Degeneration Test Flexinol Antagonist Spring Antagonist PWM-Control Summary Humanoid Finger Design Joints Tendons Friction Humanoid Finger Application Mechanics Electronics Software Regulation Finger Regulation PWM-Controlled Transformation Curve Hysteresis Delay Regulation Models by Other Authors Own Regulation Experiments Summary VI 7 Future Work Flexinol Testing Regulation Developed Finger Electronics Conclusion 101 Bibliography 107 A Code attachment 109 A.1 Software for Test Frame Control and Measurement A.2 Software for Web Surveillance of Test Frame A.3 Microcontroller Program for PWM-Control A.4 Microcontroller Program for Finger Control A.5 Computer Interface Program A.6 Interface for Microsoft Robotics Studio A.7 Matlab Scripts for Data Analysis A.7.1 Help Scripts VII VIII Chapter 1 Introduction Over the last decades, robotic technology has entered more and more areas in industry and the every day lives of humans. Robots are programmed to do complex tasks which often involve some kind of interaction with the environment. In industrial applications, a robot may be only an arm that performs a special task, or it could also be equipped with wheels that would allow it to move freely in a local environment. In such cases, the hand of the robot would often be a tool that is specially designed for a given task. However, robots that are designed to interact with humans in a physical way need human-like hands. Of course a robot could interact with humans using a stick or some other tool, but if the contact is supposed to be interpreted as human-like, hands are necessary. In health care, a robot could be a valuable assistant to a human worker, performing heavy lifts and other routine work that does not need to be done by humans alone. In such settings, a robot as adaptable and flexible as humans would be preferred, but this is still an Utopian setting. The physical adaptability of humans - our ability to use different tools to perform tasks, makes us superior to other animals. Our hands allow us to perform trivial tasks such as gripping around an unknown object while blindfolded, or to hammer in a nail. Of course, these examples depend on a well functioning regulation mechanism - our brain. 1.1 Humanoid Hands As already mentioned, a robot that is designed to interact with humans in a human-like fashion will often need hands. Robots that perform tasks in a human-like way are referred to as humanoid. Analog is a robot hand called humanoid when its design and motion is based on the principles of the human hand. Principally, there seems to be two main directions in todays research in the field of humanoid hands. The first direction has its main focus on the development of artificial hands for prosthetic applications [1, 2, 3]. These works often have criterias such as light weight, easy control, anatomical design, and in some cases, esthetics. The other branch of researchers focus more on robotic applications such as humanoid robots [4, 5, 6, 7, 8, 9, 10]. These hands have different design criterias depending on the target robot platform, are in general more complex, and possess more advanced control mechanisms than the prosthesis. Moreover, these two branches also seem to have a lot in common, such as the never ending need for adaptability. The ability of the human hand to adapt its grasp to unknown shapes and surfaces is wanted in as good as all hand designs, but is not an easy task to resolve. One state of the art humanoid robot hand is the Shadow Hand C5 [7] from The Shadow Robot Company (www.shadowrobot.com). This is a commercially developed hand and as a result, no scientific articles have been published. However, an earlier version of the hand is under research at the Bielefeld University [11]. Figure 1.1 shows three pictures of the hand in different positions. In the lower part of figure 1.1a, the actuators of the hand are shown. These are air muscles that provide light weight actuation from the forearm of the application. The hand also has a grid of touch sensors on each fingertip to provide good grasping feedback. A drawback with this design is the physical space needed for the air muscles. As the pictures clearly show, a rather large forearm is needed to fit all the muscles. 1 (a) Open hand (b) Half fist (c) Grasping an egg Figure 1.1: Shadow Hand C5 [7]. The hand is actuated by 40 pneumatic artificial muscles and has 24 degrees of freedom 1.2 Robot Hand Actuation By looking at different humanoid robot hands, it soon becomes clear that this is a field under heavy research. No standardized solutions have yet been pointed out regarding materials, actuators or design. One design criteria that seems to be commonly used is the kinematic sketch of the hand, which often is very similar to that of the human hand. However, the way that the hand is actuated varies greatly between projects. Some hands are so-called underactuated hands [1, 3, 6, 12]. This branch of hands have fewer actuators than degrees of freedom. A good example of such an underactuated hand is called the TUAT/Karlsruhe Humanoid Hand and can be found in [6, 13, 14]. The hand can be seen in figure 1.2a, and its special link mechanism is depicted in figure 1.2b. The mechanism consists of a number of link plates hierarchically connected with rods. When the actuator is used, the link plates will align such that the hand grasps around whatever object present in the hand. The TUAT/Karlsruhe Humanoid Hand is an example of the similarities between prosthetic and robotic hands, as this hand is designed to be suitable both for a humanoid service robot and for prosthetic purposes. The simple actuation of the hand is its biggest advantage in a prosthetic setting. However, this simplicity also narrows the number of robotic applications where it fits. (a) A spherical grasp around a tennis ball (b) The hand is underactuated, using only one actuator. The link mechanism distributes the grasping force over the different fingers Figure 1.2: The TUAT/Karlsruhe Humanoid Hand [6] 2 Underactuated hands depend on passive mechanisms to be able to grasp around objects. In contrast to this branch of hands are hands that have their actuators incorporated into each finger joint. However, many hybrids are also available such as the two hands depicted in figure 1.3. Both of these hands are rather large designs caused by the need for integrated motors and pumps. (a) Multi-fingered Hand for Life-size Humanoid Robots. The hand has 13 active joints and 4 passive joints, actuated with integrated servo motors [4] (b) Anthropomorphic Hand for a Mobile Assistive Robot. The hand has 8 active joints and 3 passive joints, actuated with miniature hydraulics [8] Figure 1.3: State of the art humanoid hands 1.3 Flexinol - Artificial Muscle Fibers A group of actuators not so commonly seen are artificial muscles fibers, such as Flexinol. Flexinol looks like a steel wire and has the ability to contract when a current is passed through it. Compared to its size, Flexinol is able to exert rather large forces and can be cut to any length. As the name artificial muscle fiber states, Flexinol is very similar to human muscles regarding function. Although human muscles consist of many muscle fibers, a muscle seen as a whole is very much like a Flexinol fiber. When the muscle is contracted, it is shortened and thickened, just like a Flexinol wire. Flexinol represents a kind of actuator that at first glance seems to fit perfect as actuator for a humanoid robot hand. It is small in size, very powerful and a commercial product, available at a reasonable price. Its biggest advantage is its size, that principally will allow very many Flexinol wires to be fitted inside for example a robot forehand. Although Flexinol has many clear advantages, very little research interest has been shown when it comes to using it as a robot actuator. One attempt to use Flexinol as actuator in a robot hand is proposed in [15]. The authors present a working hand, but do not mention anything regarding regulation or long term properties in the rather brief paper. Other articles about general use of Flexinol have also been found, like [16, 17, 18, 19, 20, 21, 22]. Common for most of these articles is that difficulties are uncovered, but not often solved. Some of the reported difficulties are hysteretic behaviour, high power dissipation and low strain rates. No information has been found regarding the long term properties of Flexinol. This is highly relevant information if Flexinol is to be used in a robotic application. All in all, there are several questions that need to be answered in order to determine whether Flexinol is suitable as a robot hand actuator: Does Flexinol stand the long time use as a robot hand actuator? Are the properties of Flexinol changing over time? Is it possible to regulate the contraction of Flexinol when used as actuator for a humanoid finger? 3 ˆ Is it possible to improve the strain rate of Flexinol? ˆ Is it possible to use multiple Flexinol wires in parallel? 1.4 Thesis Overview This thesis aims to answer the questions stated above. Chapter 2 contains background information about Flexinol and other actuator technologies. A brief overview of the anatomy of the human hand is given in addition to actuators and feedback sensors. Chapter 3 contains information about two of the tools used in the practical work of the thesis and chapter 4 describes the methods that were developed in order to answer the above questions. Chapter 5 contains an evaluation of the proposed methods and a discussion around regulation is found in chapter 6. Suggestions to future work is given in chapter 7 and finally, a conclusion is given in chapter 8. Following is a list over all the practical work completed during this thesis. ˆ Long term testing of Flexinol Building test frame Molding weights Design of amplifier circuit for force measurements Calibration of force and displacment transducers Design of driver circuit for Flexinol wires Programming of measurement software * Program for controlling Flexinol wires and collecting data * Web application for remote surveillance and data browsing * Matlab scripts for analyzing data ˆ PWM-control of Flexinol wire Design of microcontroller circuit with RS232 remote interface Programming microcontroller * Command interpreter for RS232 commands * Calibration algorithm * Flexinol regulation algorithm Programming computer interface * Text mode for debugging purposes * Command mode for easy operation * Continuous mode for data visualization ˆ Humanoid finger application 3D design of a humanoid finger with a torque free tendon routing scheme Printing in ABS-plastic and assembly of the finger Design, printing and assembly of radial displacement transducers Design, printing, assembly and calibration of force sensor brackets Design of micorcontroller circuit with sensor input and Flexinol driver Expansion of command set for microcontroller to include control and feedback of three wires Programming computer interface * Reuse of computer interface from PWM-testing * Interface for the Microsoft Robotics Studio framework ˆ Regulation methods Proportional regulation for PWM-control and finger joints Manual regulation of one wire using a high frequency pwm motor driver 4 1.5 Short Conclusion This thesis shows that the use of Flexinol as actuator for a robotic finger is feasible. Flexinol has many disadvantages that have to be overcome such as a limited life, hysteresis behaviour, limited strain rate and degeneration of wires but it also has advantages. Flexinol exerts very large forces compared to its own size and needs very little physical space in an application. Figure 1.4 shows a humanoid finger actuated with Flexinol wires and a test frame for testing Flexinol wires. Both products were developed during this thesis. (a) Humanoid finger developed in this thesis. The finger is actuated with three Flexinol artificial muscle fibers and controlled with a microcontroller (b) A test frame was built to investigate the long term properties of Flexinol Figure 1.4: Two products of the thesis 5 6 Chapter 2 Background In this chapter, the background theory for the thesis is presented. First, the human hand is briefly presented to later be used as motivation for the development of a humanoid finger. Secondly, available traditional actuators and actuators based on intelligent materials are presented and compared. Finally, different feedback sensors suitable for displacement- and force measurements in robotic applications are discussed. 2.1 Anatomy of The Human Hand In this section some basic anatomical principles of the human hand are presented. The human hand consists of 4 fingers and a thumb and is the main organ for physical interaction with the environment surrounding the human body. A schematic of the bones in the human hand can be seen in figure 2.1. Figure 2.1: Schematic drawing of a human hand. Carpals and Metacarpals form the wrist and palm of the hand while Proximal, Intermediate and Distal phalanges form the fingers Skeleton In figure 2.1, the carpals are known as the wrist and the metacarpals as the palm. Proximal, intermediate and distal phalanges form the three segments of the finger. The anatomy of the thumb and the wrist are 7 left out of this thesis as they are complex topics that are not needed for the presented work. Metacarpal Phalanx The metacarpal phalanx (yellow) is connected to the first finger segment (proximal phalanx) with a joint called the metacarpophalangeal joint (MCP-joint). The MCP-joint is able to perform to types of movement, flexion/extension and abduction/adduction. Flexion in this case means bending the finger while extension means extending the finger. Abduction denotes the sideways motion of the finger away from the midline of the hand. The opposite movement, adduction, means moving the finger back against the midline of the hand. Proximal Phalanx The proximal phalanx (green) is connected to the metacarpal phalanx through the MCP-joint. On the other side of the finger segment it is connected to the second finger segment (intermediate phalanx) with a joint called the proximal interphalangeal joint (PIP-joint). The PIP-joint only has one axis of motion and is therefore called a hinge joint. Flexion and extension of the PIP-joint means bending and stretching the first finger joint. Intermediate and Distal Phalanx The intermediate (blue) and distal (red) phalanges are the middle and outer segments of the finger, respectively. They are connected with the distal interphalangeal joint (DIP-joint) which is a hinge joint like the PIP-joint Tendons and Muscles The actuating mechanism of the human body is represented by muscles. However, the forces needed to grip heavy objects are so large that the muscles needed cannot be fitted inside the human hand. Instead, the muscles are placed in the forearm and the forces exerted by the muscles are transferred to the hand using tendons. This type of muscle placement is called extrinsic [23]. The efficiency of a human muscle is reported to be between 14% and 27% in the context of rowing and cycling [24]. Human Skeletal Muscles A skeletal muscle is fastened to a bone in the human body to cause movement and force exertion [24]. Thus can it be seen as an actuator for the body. The muscle itself is a bundle of single, parallel muscle fibers that are built up from muscle cells. A muscle cell consists of plates that are moved relative to each other to generate motion. The cells cause the muscle fibers and the muscle to contract when it receives a neural pulse, called an action potential. The frequency of the neural signal reception decides the contraction rate and forc
Related Search
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks