Research On Control Strategy For The Wearable Tremor Suppression Rehabilitation Robotics

As one of the most common pathological phenomenon, tremor especially pathological tremor is usually treated by drug therapy, which has not been effective enough now and has side effect sometimes. So all over the world, more and more engineering researchers are becoming interested in the field of tremor suppression, and most of them devote themselves to developing a wearable tremor suppression rehabilitation robotics, which is expected to be adjuvant to medication or a substitute for medication.But there are some difficulties lie on the research way. Firstly, as the intrinsic interaction between human and robot, it is hard to give an accurate model of the system. Secondly, the human tremor is stochastic process. Thirdly, the human tremor is usually coupled with the voluntary movement. As there are such difficulties, so it is important for us to make a reasonable control strategy, which can suppress the tremor effectively while affect the voluntary as least as possible.In this paper, we introduce a tremor-suppression strategy for the 2 DOF tremor suppression rehabilitation robotics developed by State Key Laboratory of Robotics and System. Based on the characteristics of the tremor and the magneto-rheological damper the robot use, we present a non-modeling strategy which is based upon the in-and-out of the robotics system. According to the control strategy, the magneto-rheological damper is controlled by the velocity of tremor linearly. And from the analysis results, we can find that the strategy can reduce the system's effect on voluntary movement by a considerable degree.In the signals the system gets from the sensors, the tremor signals are mixed with the signals of voluntary movement. So we need to make a distinction between tremor signals and voluntary signals to assure the execution of the strategy. And on the other hand, we can reduce the system's effect on the voluntary movement further through the tremor identification.As the frequency band of tremor and voluntary movement is usually different, we can distinguish them by filtering, but the traditional filtering algorithms have inherent delay which goes against the real-time processing requirement of the system. In this paper, we present an improved online real-time tremor pre-estimation and filtering algorithm, which is called Shifted Bandlimited Multiple Fourier Linear Combiner. The SBMFLC, which is based upon the fact that the tremor is approximately simple harmonic, is more adaptive and accurate than the past tremor filtering algorithms, so it fits the requirements of the robotic system very well.To verify the effectiveness of our strategy and filtering algorithm, we have designed an experimental tremor suppression system based on the ARM7 architecture control circuit, which includes related hardware modules, for example, main control module, input-output module, and related procedure for the control strategy and hardware. Through experiments on the tremor simulated by the vibration platform, we found that the results of the experiments were consistent with the theoretical analysis we had made before, so the effectiveness of our strategy and improvements of our algorithm in accuracy and adaptability have been verified.