Shape Memory Alloy Actuator Controller Design for Tactile Displays

Similar documents
Elements of Haptic Interfaces

Advanced Servo Tuning

2. Introduction to Computer Haptics

Haptic Perception & Human Response to Vibrations

3D Form Display with Shape Memory Alloy

A Pilot Study: Introduction of Time-domain Segment to Intensity-based Perception Model of High-frequency Vibration

Haptic Discrimination of Perturbing Fields and Object Boundaries

Servo Tuning. Dr. Rohan Munasinghe Department. of Electronic and Telecommunication Engineering University of Moratuwa. Thanks to Dr.

Experiments on the Performance of a 2-DOF Pantograph Robot Actuated by Shape Memory Alloy Wires

A Searching Analyses for Best PID Tuning Method for CNC Servo Drive

Accurate Force Control and Motion Disturbance Rejection for Shape Memory Alloy Actuators

Speech, Hearing and Language: work in progress. Volume 12

Introduction to Digital Control

Multichannel vibrotactile display for sensory substitution during teleoperation

Estimating Friction Using Incipient Slip Sensing During a Manipulation Task

High-speed wavefront control using MEMS micromirrors T. G. Bifano and J. B. Stewart, Boston University [ ] Introduction

OVEN INDUSTRIES, INC. Model 5C7-362

Biomimetic Design of Actuators, Sensors and Robots

A Tactile Sensor for Localizing Transient Events in Manipulation

Chapter 2 Introduction to Haptics 2.1 Definition of Haptics

Necessary Spatial Resolution for Realistic Tactile Feeling Display

Glossary of terms. Short explanation

Implementation of Conventional and Neural Controllers Using Position and Velocity Feedback

2B34 DEVELOPMENT OF A HYDRAULIC PARALLEL LINK TYPE OF FORCE DISPLAY

Fundamentals of Servo Motion Control

MAGNETIC LEVITATION SUSPENSION CONTROL SYSTEM FOR REACTION WHEEL

Electro-hydraulic Servo Valve Systems

Computer Assisted Medical Interventions

MEAM 520. Haptic Rendering and Teleoperation

INDIAN INSTITUTE OF TECHNOLOGY BOMBAY

Mechatronics Project Report

Figure 2: Examples of (Left) one pull trial with a 3.5 tube size and (Right) different pull angles with 4.5 tube size. Figure 1: Experimental Setup.

MEAM 520. Haptic Rendering and Teleoperation

Control Methods for Temperature Control of Heated Plates

Control design issues for a microinvasive neurosurgery teleoperator system

Introduction To Temperature Controllers

Sensors and Sensing Motors, Encoders and Motor Control

Control of the Grinding Process Using In-Process Gage Feedback

Tactile Display of Vibratory Information in Teleoperation and Virtual Environments ABSTRACT

Haptic Virtual Fixtures for Robot-Assisted Manipulation

Built-in soft-start feature. Up-Slope and Down-Slope. Power-Up safe start feature. Motor will only start if pulse of 1.5ms is detected.

Method of Accurate Countersinking and Rivet Shaving

Control and Optimization

Spatial Low Pass Filters for Pin Actuated Tactile Displays

Lecture 1: Introduction to haptics and Kinesthetic haptic devices

Peter Berkelman. ACHI/DigitalWorld

Psychophysical Characterization and Testbed Validation of a Wearable Vibrotactile Glove for Telemanipulation

Design and Controll of Haptic Glove with McKibben Pneumatic Muscle

Perception of Curvature and Object Motion Via Contact Location Feedback

Novel laser power sensor improves process control

HAPTIC INTERFACES FOR BIOMEDICAL APPLICATIONS E.P.

Vibration Feedback Models for Virtual Environments

Basic Tuning for the SERVOSTAR 400/600

Performance Issues in Collaborative Haptic Training

Cutaneous Feedback of Fingertip Deformation and Vibration for Palpation in Robotic Surgery

Understanding PID Control

Introduction to Servo Control & PID Tuning

Hybrid Input Shaping and Non-collocated PID Control of a Gantry Crane System: Comparative Assessment

A Machine Tool Controller using Cascaded Servo Loops and Multiple Feedback Sensors per Axis

GLOSSARY OF TERMS FOR PROCESS CONTROL

Development of a telepresence agent

MEM380 Applied Autonomous Robots I Winter Feedback Control USARSim

Sensors and Sensing Motors, Encoders and Motor Control

Ball Balancing on a Beam

UNIVERSITY OF JORDAN Mechatronics Engineering Department Measurements & Control Lab Experiment no.1 DC Servo Motor

Real-Time Bilateral Control for an Internet-Based Telerobotic System

Improving Telerobotic Touch Via High-Frequency Acceleration Matching

6.4 Adjusting PID Manually

LAB 1 AN EXAMPLE MECHATRONIC SYSTEM: THE FURBY

STANDARD TUNING PROCEDURE AND THE BECK DRIVE: A COMPARATIVE OVERVIEW AND GUIDE

Position Control of a Hydraulic Servo System using PID Control

The Haptic Impendance Control through Virtual Environment Force Compensation

Continuous Rotation Control of Robotic Arm using Slip Rings for Mars Rover

INTELLIGENT ACTIVE FORCE CONTROL APPLIED TO PRECISE MACHINE UMP, Pekan, Pahang, Malaysia Shah Alam, Selangor, Malaysia ABSTRACT

Magnetic Levitation System

Active sway control of a gantry crane using hybrid input shaping and PID control schemes

Chapter 2 Mechatronics Disrupted

LDOR: Laser Directed Object Retrieving Robot. Final Report

Experiment 9. PID Controller

G320X MANUAL DC BRUSH SERVO MOTOR DRIVE

V2018 SPINSTAND AND NEW SERVO-8 SYSTEM

Design and Operation of a Force-Reflecting Magnetic Levitation Coarse-Fine Teleoperation System

Vibrotactile Device for Optimizing Skin Response to Vibration Abstract Motivation

The Discussion of this exercise covers the following points: Angular position control block diagram and fundamentals. Power amplifier 0.

Lecture 7: Human haptics

VIBROTACTILE FEEDBACK FOR INDUSTRIAL TELEMANIPULATORS

Technical Cognitive Systems

Integrating Tactile and Force Feedback with Finite Element Models

Procidia Control Solutions Dead Time Compensation

Think About Control Fundamentals Training. Terminology Control. Eko Harsono Control Fundamental - Con't

REDUCING THE STEADY-STATE ERROR BY TWO-STEP CURRENT INPUT FOR A FULL-DIGITAL PNEUMATIC MOTOR SPEED CONTROL

This is a postprint of. The influence of material cues on early grasping force. Bergmann Tiest, W.M., Kappers, A.M.L.

Robot Sensors Introduction to Robotics Lecture Handout September 20, H. Harry Asada Massachusetts Institute of Technology

Design of Cylindrical Whole-hand Haptic Interface using Electrocutaneous Display

Rotary Motion Servo Plant: SRV02. Rotary Experiment #02: Position Control. SRV02 Position Control using QuaRC. Student Manual

ECE 203 LAB 6: INVERTED PENDULUM

Lecture 9: Teleoperation

WEARABLE HAPTIC DISPLAY FOR IMMERSIVE VIRTUAL ENVIRONMENT

Miniaturization of a Quasi-Servo Valve and Its Application to Positon Control of a Rubber Artificial Muscle with Built-in Sensor

A Feasibility Study of Time-Domain Passivity Approach for Bilateral Teleoperation of Mobile Manipulator

Transcription:

34th IEEE Conference on Decision and Control New Orleans, Dec. 3-5, 995 Shape Memory Alloy Actuator Controller Design for Tactile Displays Robert D. Howe, Dimitrios A. Kontarinis, and William J. Peine Division of Applied Sciences, Harvard University Cambridge, MA 38 howe@das.harvard.edu Abstract This paper describes the design of a controller for a tactile shape display for dexterous telemanipulation applications. This device consists of an array of pin elements which rest against the human finger tip. Shape memory alloy (SMA) wire actuators raise individual pins to approximate the desired surface shape on the skin. To overcome the slow response times of SMA actuators we have implemented a feedforward derivative compensator and pneumatic cooling. The hysteretic and nonlinear response of the SMA actuators required the use of a closed loop controller with position feedback using an optical emitter-receiver pair. This also improves robustness to the load variations which result from human interaction. Introduction In teleoperated manipulation systems, a human operator controls a robot located in a distant, hazardous, or difficult to reach worksite. These systems have been used for many years in the nuclear industry and for submarine exploration, and new applications are emerging in extravehicular activities in space and minimally invasive surgical procedures (Sheridan 99). Most attention has focused on manipulation at the arm level, although recent work on multi-fingered hands promises to increase dexterity in complex and delicate manipulation tasks. Because humans rely on tactile information to control these types of manipulation tasks (Johansson and Westling 984), tactile feedback from the remote robot to the human operator is essential for good performance. Our research is aimed at developing this tactile feedback capability. We are working to identify the physical parameters that must be sensed at the remote manipulator, to develop sensors to measure these parameters, and to build displays to convey them to a human operator. In our tactile shape relay system, the shape of an object in contact with the remote robot's finger tip is measured by a tactile array sensor. This shape is recreated on the human operator's finger tip by a tactile shape display. This device consists of an array of pin elements or "tactors" which are raised and lowered to approximate the desired shape. Further details of the design and performance the overall shape relay system, with results of experimental tests in telemanipulation tasks, can be found in (Kontarinis et al. 995). In this paper we present the controller design for the tactile shape display device. Our goal was to develop a tactile shape display for use in the Figure. Shape display and human finger tip.

V LEVER SMA WIRE DETECTOR PIN PIVOT SPRING EMITTER Figure. Side view of one actuator. grasping surface of a force-reflecting teleoperated master robot hand. This poses extremely difficult design challenges. First, the display must be small enough that it fits between the fingers when manipulating an object, and light enough to avoid loading the master and limiting force reflection range and responsiveness. In addition, because it is located at the point of contact between the manipulator and the operator's finger tip, it must be strong enough that the entire contact force can be supported by the display elements while maintaining the desired shape. Finally, the display's spatial and temporal bandwidth should approach the capabilities of the human cutaneous system. Based on a survey of psychophysical and neurophysiological data (e.g. Johansson and Vallbo 983, Loomis and Lederman 986), we selected design goals of approximately mm center-to-center element spacing, 3 mm total vertical excursion, Hz bandwidth, and at least N force per tactor. Mechanical Design The shape display raises pins against the human finger tip skin to approximate the desired shape (Figure ). We selected shape memory alloy (SMA) wires as actuators because of their very high force-to-volume and force-to-weight ratios. The mechanical design of one element of the shape display is shown in Figure. A length of SMA wire is attached to a rigid frame at one end and to a small lever at the other. A spring connected between the lever and the frame keeps the wire in tension and provides a restoring force. The SMA wires are actuated by heating with an electric current. The elevated temperature results in a material phase change which increases the tension and/or shortens the length between the ends of the wire. This causes the lever to pivot about a fixed shaft. The other end of the lever then forces up a pin which rests against the tip of the operator's finger. Force Newtons...8.6.4...4.6.8...4.6.8 Current (A). Figure 3. Force versus current for triangle wave inputs of. Hz (-),. Hz (..), and Hz (- -). The levers provide a 3: reduction in force and amplification in displacement. The wires we use are 3 mm long and.75 mm in diameter, with a 9 C critical temperature. The shape display consists of four layers, slightly offset, each layer having six actuators. The center-to-center spacing of the tactors is approximately. mm. 3 SMA Characteristics SMA actuators are difficult to accurately control due to long thermal time constants, hysteresis, and other nonlinear characteristics. To illustrate these difficulties and to motivate our controller design, we describe a basic performance measurement for the SMA wires used in our display. We measured isometric forces as a function of input current directly at the lever of the display. Figure 3 shows the output force generated by applying triangle wave current inputs to the wire. The solid curve represents a. Hz triangle wave current input (i. e. 5 sec ramp up and 5 sec ramp down). At this slow time scale the wire is almost in thermal equilibrium at each instant, so this represents the "steady state" dependence of force on current. Hysteresis of -% of the full force range is evident in this curve; this hysteresis is intrinsic to the phase change which produces the shape memory effect used for actuation. The other curves, which represent. and. Hz frequencies, show an increasing phase lag. This is due to the slow thermal response of the SMA wires. There is a significant delay between the increase of the input current and the appearance of the force increase due to the integrating effect of the thermal mass of the wire. On the descending phase, the delay is due to the slow cooling rate, so that the temperature drops more slowly than the current. Results for free motion

Input Command Output Displacement Open Loop Output Displacement Controller 5 Time (seconds).5 mm Figure 4. Time response improvement with feedforward compensation and air cooling. (zero force) are similar. These effects necessitate the use of appropriate control schemes to achieve the desired performance. 4 Bandwidth Considerations Our first consideration was the need to improve response times. Our compensator is based on a simple thermal model of the wire. Note that the control input is unilateral in this case: we can only heat the wire with electrical current, while cooling is due to almost exclusively to convective heat transfer to surrounding air. This requires the use of different strategies for increasing and decreasing commands. Based on the thermal model, a proportionalderivative (PD) feedforward control law with a large derivative term has been implemented. This scheme introduces large current peaks at rapid positive changes in the command to heat the wires quickly. The duration of the peak is critical since a long high current input to the wire will lead to overshoot and even damage to the wire by overheating. To heat the wire at a desired temperature both the amplitude and the duration of the "overload" current must be modulated to compensate for the new heat energy state reached at every servo cycle. As a first order compensation for the changing initial conditions in heat, the input current to the wire is modulated by adding a weighted average of the previous and present inputs. The feedforward compensation can be described as I t out [ ] = K F + K Ḟd + K p d v old I out [ t t] Actual Position (mm).5.5.5.5 Desired Position (mm) Figure 5. Hysteresis in displacement response to a triangle wave input with feedforward compensation and air cooling. where Iout[t] is the current input to the wire, Iout[t-Dt] is the previous input to the wire, Fd is the desired force, Kp is a proportional gain, Kv is a velocity gain and Kold is a weighting factor that averages the present and previous commands. This control law results in a rapid response to a step command (derivative term) with a geometric decay (weighted average term), which is tuned to the thermal mass of the wire. The feedforward compensation also decreases the response time to decreasing commands by diminishing the current at a faster rate due to the derivative term. However, the response to decreasing commands is fundamentally limited by the rate at which heat leaves the wire. Decreasing this time constant requires a higher convective cooling rate. For this purpose we have added small pneumatic jets which force air across the wires. Based on the thermal model, the cooling rate and derivative feedforward compensation can be balanced to produce a quasi-linear system with symmetric response to rising and falling commands. Even faster response can be obtained by liquid cooling, at the expense of a more complex mechanical design. Figure 4 shows the performance improvement that results from use of the feedforward derivative compensator and pneumatic cooling. Step response rise time ( to 63%) decreases from to 65 ms, and fall time from 5 to 65 ms. 5 Hysteresis Considerations These bandwidth compensation techniques do not eliminate hysteresis, as illustrated by the response 3

Actual Position (mm).6.4..8.6.4...4.6.8..4.6.8 Desired Position (mm) Figure 6. Response to a triangular wave input with position feedback. to a triangle wave input shown in Figure 5. Previous work on SMA control has shown that it is possible to estimate the phase state of the SMA material through measurements of electrical resistance (Ikuta et al. 988), which can be used in the design of a controller to minimize hysteresis effects. However, state estimation alone is not sufficient in our application. As the material phase state changes, the effective mechanical modulus and rest length of the wire changes. Thus the phase information specifies the relationship between the length of the wire and the force on the wire, but does not provide a direct measurement of either parameter. The force on the pin elements of the display varies with a number of factors in this application, including the human operator's finger tip skin stiffness, the number of adjacent pins that are raised, and how hard the operator is pushing against the display. Thus direct sensing of either force or displacement is required for good control. We have elected to sense displacement, as this is the output parameter in most uses of the shape display. The position sensors are infrared LEDs and phototransistors directed at the bottom each lever (Figure ). As the lever rises, the light reflected back into the phototransistors decreases. The relation between the sensor output and the displacement of the lever has been determined by an independent calibration. As an initial test of the effectiveness of position feedback in eliminating hysteresis and load dependence, we have used the sensed position Displacement (mm).6.4..8.6.4. -. 3 4 5 6 7 8 9 Time (seconds) Figure 7. Tracking response with a 7.5 g load with feedforward compensation, air cooling, and position feedback. signal in a simple proportional position control loop. The addition of position feedback greatly reduces the hysteresis associated with the SMA wires, as shown in Figure 6. In addition, this control scheme reduces the variability between elements (and thus the need to calibrate each element of the display), and decreases sensitivity to variations in human finger tip stiffness. As a demonstration of the display's ability to follow trajectories under load, we placed a 7.5 g mass on one lever and recorded the response to a triangle wave. Figure 7 shows that even this simple controller results in improved performance in the loaded case. To further improve performance in the loaded case, we have designed a model-based controller which incorporates both SMA and human load terms. The SMA portion includes material phase state estimates based on electrical resistance measurements. This permits inclusion of the effective stiffness and rest length of the wire in calculations of the required actuator force. The human load portion of the model is based on measurements of the mechanical properties of the human finger tip skin. In addition to improving force calculations, the human skin model is used to compensate for skinmediated interaction between adjacent actuator elements. 6 Conclusions The techniques used to improve control of the SMA actuators include: 4

Faster heating response by feedforward derivative compensation; Faster cooling response by pneumatic cooling; Reduced hysteresis through the use of position sensing. These improvements result in a bandwidth with a -3dB point between 6 and 7 Hz (Figure 8). Ongoing work is directed at further raising bandwidth to the desired Hz value through liquid cooling, and implementing the modelbased controller using state properties of the SMA material and human skin mechanics. Acknowledgments Funding for this work was provided by the Office of Naval Research under ONR Grants No. N4-9-J-887 and N4-9-J-84. Amplitude (db) - -4-6 -8 - - -4 5 5 Frequency (Hz) Figure 8. Frequency response with feedforward compensation, air cooling, and position feedback. References Ikuta, K., Tsukamoto, M., and Hirose, S., 988, "Shape Memory Alloy Servo Actuator System with Electric Resistance Feedback and Application for Active Endoscope," in Proceedings of 988 IEEE International Conference on Robotics and Automation, pp. 47-43. Johansson, R.S., and Vallbo, A.B., 983, "Tactile sensory coding in the glabrous skin of the human hand," Trends in NeuroSciences, Vol. 6, No., pp. 7-3. Johansson, R.S., and Westling, G., 984, "Roles of glabrous skin receptors and sensorimotor memory in automatic control of precision grip when lifting rougher or more slippery objects, Experimental Brain Research, Vol. 56, pp. 55-564. Kontarinis, A., et al., "A Tactile Shape Relay System for Telemanipulation," Proc. 995 IEEE International Conference on Robotics and Automation, Nagoya, Japan, May 995, p. 64. Loomis, J. M., Lederman, S. J., 986, "Tactual Perception," Handbook of Human Perception and Performance, K. B. Boff, L. Kaufman, and J. Thomas (Eds.), Ch. 3, New York, Wiley. Sheridan, T. B., 99, Telerobotics, Automation, and Human Supervisory Control, Cambridge, MIT Press. 5