Thin film shape memory alloys (SMAs) have the potential to become a primary actuating mechanism for mechanical devices with dimensions in the micron-to-millimeter range requiring large forces over long displacements. The work output per volume of thin film SMA microactuators exceeds that of other microactuation mechanisms such as electrostatic, magnetic, thermal bimorph, piezoelectric, and thermopneumatic, and it is possible to achieve cycling frequencies on the order of 100 Hz due to the rapid heat transfer rates associated with thin film devices. In this paper, a quantitative comparison of several microactuation schemes is made, techniques for depositing and characterizing Ni-Ti-based shape memory films are evaluated, and micromachining and design issues for SMA microactuators are discussed. The substrate curvature method is used to investigate the thermo-mechanical properties of Ni-Ti-Cu SMA films, revealing recoverable stresses up to 510 MPa, transformation temperatures above 32/spl deg/C, and hysteresis widths between 5 and 13/spl deg/C. Fatigue data shows that for small strains, applied loads up to 350 MPa can be sustained for thousands of cycles. Two micromachined shape memory-actuated devices-a microgripper and microvalve-also are presented.

Thin film shape memory alloy microactuators

One important challenge of microsystems design is the implementation of miniaturized actuation principles efficient at the micro-scale. Shape memory alloys (SMAs) have early on been considered as a potential solution to this problem as these materials offer attractive properties like a high-power to weight ratio, large deformation and the capability to be processed at the micro-scale. This paper reviews various attempts made to introduce SMAs in microsystems as well as design principles for SMA microactuators. It presents the status of current research and potential developments from a material research perspective.

Shape memory alloys for microsystems: A review from a material research perspective

Spinal cord and other lo cal injuries ofte n lea d to partial paralysis while the brain stays fully functional. When this partial paralysis occurs in the hand, these individuals are not able to execute daily activities on their own even if their a rms are func tional. To remedy this problem, a lightweight, low-profile orthotic exoskeleton has been designed to restore dexterity to paralyzed hands. The exoskeleton’s movements are controlled by the user’s available electromyography (EMG) signals. The device has two actuators controlling the index finger flexion that can be used to perform a pinching motion against a fixed thumb. Using this orthotic device, a new control technique was developed to allow for a natur al reaching and pinching sequence by utilizing the natural residual muscle activation patterns. To design this controller, two actuator control algorithms were explored with a qua driplegic (C5/ C6) subject and it was determined that a simple bina ry c ontrol algorithm allowed for faster interaction with objects over a variabl e cont rol algorithm. The binary algorithm was then used as an enabling algorithm to activate the exoskeleton movements when the natural sequence of muscle activities found a pattern related to a pinch. This natural pinching technique has shown significant promise toward realistic neural control of wearable robotic devices to assist paralyzed individuals.

/pdf/an-emg-controlled-hand-exoskeleton-for-natural-pinching-3zow4td7p9.pdf

An EMG-Controlled Hand Exoskeleton for Natural Pinching

To restore dexterity to paralyzed hands, we have designed and constructed a lightweight, low-profile orthotic exoskeleton controlled by the user's residual electromyography (EMG) signals. In this paper, we compared several simple strategies to control the orthotic device for a quadriplegic (C5/C6) subject. When contralateral arm control was employed, we found that a simple on/off strategy allowed for faster interaction with objects, while variable control provided more controlled interactions, especially with deformable objects. Furthermore, we designed a control strategy that allowed for a natural reaching and pinching sequence without the use of the contralateral arm. We validated that the EMG signal from the ipsilateral biceps could be used to develop an extremely reliable natural reaching and pinching algorithm. This evaluation showed that different control strategies may be appropriate for different situations, and further investigation on the natural algorithm is crucial.

/pdf/comparison-of-control-strategies-for-an-emg-controlled-d3phq6hgg1.pdf

Comparison of control strategies for an EMG controlled orthotic exoskeleton for the hand

This paper reports the prosthetic hand controller discriminating ten forearm motions from two channels of EMG signals. The controller uses the real-time learning method that is defined as simultaneously controlling a prosthetic hand and learning to adapt to the operator's characteristics. In this method, the controller is divided into three units. The analysis unit extracts useful information for discriminating motions from EMG. The adaptation unit learns the relation between EMG and control command and adapts to the operator's characteristics. The trainer unit generates training data and makes the adaptation unit learn in real-time. In experiments, the proposed controller performs discriminating a maximum of ten forearm motions including four wrist motions and six hand motions. In an eight forearm motions experiment, the five subjects' average discriminating rate, which serves an index of accurate controlling, was 85.1%. Two groups occur from this result, one marks a high performance (91.7%) and another does not (75.2%). The paper discusses the factors of this difference in performance from both phases of training and reasons that the low proficiency leads to undesirable results in the latter groups. Besides, in the ten forearm motions experiment the average discriminating rate of three subjects who achieve high performance in the previous experiment was 91.5%. This paper concludes that the effectiveness of the real-time learning method is confirmed by these experiments.

EMG prosthetic hand controller using real-time learning method

The electromyographic (EMG) signal from active muscle is one of the most effective signals for controlling externally powered upper extremity prostheses. However, the EMG signal depends on physical condition, the state of mind, and so on, so it is difficult to use the original EMG signal to control an externally powered upper extremity prosthesis directly. A discriminating system using an integral type of EMG signal processing and a neural network is proposed for such an application. The neural network is used to learn the relation between the integral values of the EMG signals and the performance desired by the handicapped person. It has been found that total discrimination time can become shorter than Fourier transform processing and the discrimination system can discriminate seven performances from the EMG signals with a probability of 95.5% using integral processing of the EMG signal.

A discrimination system using neural network for EMG-controlled prostheses-Integral type of EMG signal processing

A micron-sized arm which has large work space and uses reversible TiNi thin film is proposed. The transformation characteristics as analyzed by the DSC method shows that RTiNi can be controlled around 60/spl deg/C. The design method and fabrication process for the arm are described, and the arm was fabricated on silicon wafer. It was shown that micron sized arm using three RTiNi thin films could be fabricated based on the trial fabrication of simple micron sized arm using a RTiNi thin film.

Trial fabrication of micron sized arm using reversible TiNi alloy thin film actuators

The robot finger force sensor needs tobe designed small in order to fit it in the top part of the robot finger. However, because the finger force sensor of small size has little possibility to be designed as complex structure, it is difficult to produce non-interacting signals from the sensor. In this paper, a trial production of the simple structure of a robot finger force sensor using a capacitance detector is described. Besides, a neural nework was used for processing interacting signals. Then, though there are a series interactions between each signal of output voltages and nonlinearity in this sensor, it is shown that the sensor acted effectivety as a robot finger force sensor. In addition, as compared with a conventional method using as inverse matrix, this new method is more accurate.

A new robot finger force sensor using neural network

So far, the researches on SMA (= shape memory alloy) thin film micro actuator for micro machine were reported about the metallurgical aspect, the sputter deposit process and the fabrication of devices with the SMA thin film. However the properties of SMA thin film as servo actuator has not been studied. On the other hand, the several characteristics of bulk SMA wire actuators to control it had been studied. In this paper, based on the same aspect as the bulk SMA wire actuator, the dynamical properties, the mathematical model and control of SMA thin film actuator were investigated experimentally. As the results, the summary is as follows. 1) The mathematical model of output force of SMA thin film actuator is expressed as the nonlinear and continuous function of temperature and strain. 2) The above function is found linearizable. Thus the linear mathematical model was obtained. 3) Tensile force could be controlled by the force sensor feedback system of the SMA thin film actuator.

S. Shimizu

Papers

A discrimination system using neural network for EMG-controlled prostheses-Integral type of EMG signal processing

Trial fabrication of micron sized arm using reversible TiNi alloy thin film actuators

A new robot finger force sensor using neural network

Mechanical Properties and Control of Shape Memory Alloy Thin Film Actuator.

A millimeter-Sized Robot Using SMA and Its Control