Qing Xu, M.S.
Jian Q. Sun PhD, Tariq Rahman PhD, Ken Barner PhD
Department of Mechanical Engineering
University of Delaware
Newark, DE 19716
INTRODUCTION: Tremor is a rhythmic involuntary oscillatory movement of body parts, with a relatively fixed frequency and amplitude. On the basis of their appearance, tremor has been classified in two broad groups as resting tremor, which appears at rest in fully supported limbs, and action tremor, which occurs during movement and posture. Tremor can be observed in healthy subjects as well as in patients with various diseases. It is estimated that there are more than 5 million people, over 40, in the United States affected by tremor. Physiological tremor is present in normal subjects and its frequency generally falls in the range of 8-12Hz, while pathological tremor which results from diseases such as Parkinson s is in a lower frequency range. In many cases, tremor can be severe enough to make daily activities such as eating, drinking and writing very difficult.
Drug treatment of tremor is available today, but there is no standard reliable prescription and undesirable side effects are usually unavoidable. Because of the difficulties with drug treatment, it is necessary to develop assistive devices to help people with tremor improve the quality of their life. In this abstract, we present two approaches for tremor suppression, one of which is signal processing approach and the other is force feedback approach.
MATERIALS & METHODS: In the signal processing approach, an adaptive finite impulse response (FIR) filter is developed for reducing the tremor in the output signal of an assistive device such as a joystick for writing and for wheelchair. The adaptive FIR filter is able to learn about the tremor signal and eventually to produce a signal that cancels the unwanted component of the arm motion. The least-mean-square (LMS) algorithm is used for updating the filter. This approach can handle tremor with multiple frequencies and with slowly time varying characteristic. Because it is parallel to the main signal flow, the filter introduces no time delay to the voluntary movement signal, which is a major problem with other methods for tremor suppression. Another advantage of the adaptive filter is that no information other than the frequency range of tremor has to be known beforehand.
The second method is force feedback control. As the name suggests, this approach uses force to cancel out the tremor in the arm motion. The patient would feel resistance from the controller, certain kind of resistance, while in the signal processing case, the patient would feel no resistance and only see the effect of the filter on the screen. The force feedback approach is implemented on a PerForce hand controller manufactured by Cybernet, Inc. The hand controller is a force reflecting device originally designed for the space station. Several force profiles are available and we choose the force to be proportional to the velocity of the handle. Thus the effect is equivalent to increasing the damping of the system. This is called active damping and is more flexible than the usual mechanical passive damping. One important accomplishment in this approach is system modeling, which is the basis of design for controllers.
RESULTS: Experiments have been done with both approaches. PerForce hand controller was used as the input device in both cases. Human arm tremor was created by attaching a 4 Watt DC motor on the person s hand. The frequency of the created tremor was around 8Hz. Figure 1 shows a typical set of writing tests with adaptive FIR filter. It can be seen that the filter effectively removes the tremor in the hand movement. It can also be seen that the learning process of the adaptive filter is so short that the transient response can hardly be seen. Figure 2 shows the writing tests with force feedback. It is clear that by increasing the damping of the system, tremor can be significantly reduced.
Figure 1. Writing test with adaptive FIR filter. Left: before the filter. Right: after the filter.
Figure 2. Writing test. Left: with force feedback. Right: without force feedback
CONCLUSION: The adaptive FIR filter proves to be satisfactory in eliminating tremulous movement signal of human arms. Unlike the classic structure of serial filtering, parallel filtering theoretically causes no delay to the voluntary movement. The LMS algorithm and the low order of the FIR filter gives a quick learning adaptive filter with simple computation. The high performance of the FIR filter shows that it can be used as a component in various assistive devices such as orthosis and power wheelchair. The force feedback approach shows the effect of increasing damping in tremor suppression. A mass-damping system is equivalent to a low pass filter. Although low pass filter has some deficiencies for tremor suppression, such as time delay, it is effective in reducing severe tremor. Compared with the mechanical damping, the force- feedback controller is more flexible. The methodology developed in this work for tremor suppression can be extended to deal with other human movement disorders.
[Return to Table of Contents, Back to Research Symposium Home Page, Orthopaedic Department Home Page]