The slacking behaviour or lack of participation from impaired patients during robotic rehabilitation therapy is one of the factors that slow down their recovery. The implementation of Assist As Needed (AAN) control law in the robotic assisted rehabilitation treatment may alleviate this problem and encourage the patients to be actively involved in the rehabilitation exercises. This paper presents a new Assist As Needed control strategy for an upper limb rehabilitation robot in assisting subjects with various levels of capabilities to regain their original upper limb’s functionality in realizing basic motions in eating activity. The controller consists of Proportional, Integral, Derivative (PID) controller in the feedback loop, with an adjustable gain K that varies according to the user’s level of capability. A Force Sensing Resistor (FSR) is used to identify the user’s upper extremity capability level. The controller regulates the necessary amount of assistance provided by the robot based on the information obtained from the sensor. The automatic adjustment of the robot’s assistance to the subjects leads them to put in their own effort in accomplishing the desired movements. The proposed control strategy is simple, easy to program, and mathematically less complicated. A prototype of the wearable upper limb rehabilitation robot has been built and a Graphical User Interface (GUI) has been developed using MATLAB software to facilitate the rehabilitation process and for progress monitoring. The simulation and experimental results have proven that the proposed control strategy is successful in regulating the necessary amount of robot assistance according to the patients’ level of capability. The proposed controller has effectively driven the upper limb rehabilitation robot to achieve the desired trajectory with zero steady state error, percentage overshoot less than 8% and settling time below 6 seconds, whilst providing the correct amount of robotic assistance in accordance to the subjects’ capability level.
ABSTRAK: Reaksi kurang respon dari pesakit kurang keupayaan semasa terapi pemulihan robotik adalah satu faktor melambatkan kadar pemulihan. Pelaksanaan teknik kawalan Bantu Apabila Diperlukan (AAN) dalam rawatan pemulihan dengan bantuan robot dapat membantu dan mendorong pesakit terlibat secara aktif dalam latihan pemulihan. Artikel ini membentangkan strategi kawalan baru, iaitu Bantu Apabila Diperlukan oleh robot pemulihan bagi anggota atas pesakit yang mempunyai pelbagai tahap kemampuan, dalam mengembalikan fungsi asas gerakan tangan seperti aktiviti makan. Teknik kawalan terdiri daripada kawalan Berkadar, Integral, Terbitan (PID) dalam lingkaran tindak balas, dengan pemboleh ubah K mengikut tahap kemampuan pesakit. Alat pengukur Resistan Daya Rasa (FSR) digunakan bagi mengenal pasti tahap kemampuan maksima pesakit dalam menggerakkan tangan. Berdasarkan maklumat yang diperoleh daripada sensor, teknik kawalan akan menghantar maklumat kepada robot bagi membantu pesakit. Bantuan automatik yang dibekalkan robot kepada pesakit akan mendorong pesakit berusaha melakukan gerakan yang diperlukan. Strategi kawalan yang dicadangkan ini adalah ringkas, mudah diprogramkan dan kurang rumit dari segi matematik. Sebuah prototaip robot pemulihan anggota tangan telah dibina dan sebuah platform grafik bagi pengguna (Antara Muka Grafik Pengguna, GUI) telah dibangunkan menggunakan perisian MATLAB bagi memudahkan proses pemulihan dan pemantauan kemajuan pesakit. Hasil simulasi dan eksperimen membuktikan bahawa strategi cadangan kawalan ini berjaya mengatur jumlah bantuan daripada robot bersesuaian dengan tahap kemampuan pesakit. Teknik kawalan yang dicadangkan telah berjaya menggerakkan robot pemulihan tangan bagi mencapai lintasan gerakan yang diinginkan dengan ralat sifar pada keadaan stabil, peratusan ayunan berlebihan kurang daripada 8%, masa penyelesaian bawah 6 saat dan pada masa sama, memberikan maklumat bantuan robot yang tepat, bersesuaian dengan tahap kemampuan pesakit.
Multi-feature data fusion based on common space model and recurrent convolutional neural networks for EEG tristimania recognition used in upper limb rehabilitation exercises
