MLP ve SVM Makine Öğrenme Modelleri Kullanarak Robotik El Kontrolü

Öz

Biyolojik sinyallere dayalı robotik el kontrol sistemleri hareket analizinin doğruluğunu, sağlamlığını ve doğallığını iyileştirmeyi amaçlamaktadır. Makalede parmak hareketlerinin sınıflandırılmasını ve öznitelik değişkenlerini kullanarak hareket tahmini ile robotik el kontrolü konu edinilmektedir. Çoklu DOF parmak kontrolü için makine öğrenme modellerinin karşılaştırması ve hibrit model veri toplama sistemi kullanılarak oluşturulan veri setinin eğitilmesi ile modellerin performans değerlerinin yükseltilmesi sağlanmaktadır. Geliştirilen modeller, belirli ampute birey kullanıcılarının etkili bir şekilde robotik el kullanabilmesi için özel olarak tasarlanmış çoklu DOF kontrolünde kullanılmaktadır. Gelişmiş yöntemlerle makine öğrenme modellerinin tahmin çıktıları kullanılarak servo motorların hareketleri sağlanmaktadır. Doğruluk oranları sırasıyla %99,15 ve %97,47 olan kontrol modelleri için MLP ve SVM makine öğrenmesi algoritmaları kullanılmaktadır. Veri setine haricen uygulanan her bir hareketin vektörel filtrelenmiş hareket verisi, geliştirilen modelerin çalışması ile robotik el hareketleri sağlanmaktadır. Her bir parmaktan metal dişli servo motor ile el bileğine doğru kapanma ve el bileğinden dışa doğru açılma gibi 0-128 derece hareket aralığında parmak hareketleri geliştirilen interpolasyon kontrol yöntemi ile sağlanmaktadır.

Anahtar Kelimeler

• sEMG
• makine öğrenmesi
• interpolasyon

Robotic Hand Control Using MLP and SVM Machine Learning Models

Abstract

Robotic hand control systems based on biological signals aim to improve the accuracy, robustness and naturalness of motion analysis. In the article, the classification of finger movements and the provision of robotic hand control with motion estimation using attribute variables are discussed. It is ensured that the performance values of the models are increased by comparing the machine learning models for multi-DOF finger control and training the data set created using the hybrid model data collection system. The developed models are used in multi-DOF control, which is specially designed for certain amputees users to use robotic hands effectively. The movements of servo motors are provided by using the prediction outputs of machine learning models for multi-DOF control with advanced methods. MLP and SVM machine learning algorithms are used for control models with accuracy rates of 99.15% and 97.47%, respectively. Vector filtered motion data of each movement applied externally to the data set is provided with robotic hand movements with the operation of the developed models. Finger movements in the range of 0-128 degrees of motion, such as closing towards the wrist and opening outward from the wrist with a metal toothed servo motor from each finger, are provided by the developed interpolation control method.

Keywords

• sEMG
• machine learning
• interpolation

Kaynakça

1. [1] Uguz, S. Makine ögrenmesi-Teorik yönleri ve Python uygulamaları ile bir yapay zeka ekolü. Nobel Yayınları, 2019.
2. [2] Guo, S., Pang, M., Gao, B., Hirata, H., & Ishihara, H. Comparison of sEMG-based feature extraction and motion classification methods for upper-limb movement, Sensors. 2015; https://doi.org/10.3390/s150409022
3. [3] Lima, A. A. M., Araujo, R. M., dos Santos, F. A. G., Yoshizumi, V. H., de Barros, F. K., Spatti, D. H., ... & Dajer, M. E. Classification of hand movements from EMG signals using optimized map, International Joint Conference on Neural Networks (IJCNN). 2018; https://doi.org/10.1109/IJCNN.2018.8489150
4. [4] Bilgin, M., Makine Öğrenmesi Teorisi ve Algoritmaları. Papatya Bilim, 2022, İstanbul.
5. [5] Mersinkaya, I., & Kavsaoglu, A. R. A Data Acquisition System with sEMG Signal and Camera Images for Finger Classification with Machine Learning Algorithms. Engineering, Technology & Applied Science Research. 2024; https://doi.org/10.48084/etasr.7040
6. [6] Meattini, R., Benatti, S., Scarcia, U., De Gregorio, D., Benini, L., & Melchiorri, C. An sEMG-based human-robot interface for robotic hands using machine learning and synergies. IEEE Transactions on Components, Packaging and Manufacturing Technology. 2018; https://doi.org/10.1109/TCPMT.2018.2799987
7. [7] Ishii, C., Harada, A., Nakakuki, T., & Hashimoto, H. Control of myoelectric prosthetic hand based on surface EMG. IEEE International Conference on Mechatronics and Automation. 2011; https://doi.org/10.1109/ICMA.2011.5985757
8. [8] Ricardez, G. A. G., Ito, A., Ding, M., Yoshikawa, M., Takamatsu, J., Matsumoto, Y., & Ogasawara, T. Wearable device to record hand motions based on EMG and visual information. 14th IEEE/ASME International Conference on Mechatronic and Embedded Systems and Applications (MESA). 2018; https://doi.org/10.1109/MESA.2018.8449178

9. [9] Semprini, M., Cuppone, A. V., Delis, I., Squeri, V., Panzeri, S., & Konczak, J. Biofeedback signals for robotic rehabilitation: assessment of wrist muscle activation patterns in healthy humans. IEEE Transactions on Neural Systems and Rehabilitation Engineering. 2016; https://doi.org/10.1109/TNSRE.2016.2636122
10. [10] Lu, M. H., Huang, C. C., Lin, C. L., & Shen, Y. J. Demonstrative DC Motor Control under Power Communication Network. Sensor Netw Data Commun. 2016; https://doi.org/10.4172/2090-4886.1000139
11. [11] Mufazzal, S. Studies on Grübler’s Formula for Mobility and Connectivity. In: Syed Wasiul Hasan Rizvi, editor. Innovative Product Design & Development. India; 2017. pp. 141-145.
12. [12] Du, K.-L., Leung, C.-S., Mow, W. H., & Swamy, M. N. S. Perceptron: Learning, Generalization, Model Selection, Fault Tolerance, and Role in the Deep Learning Era. Mathematics, 2022; https://doi.org/10.3390/math10244730
13. [13] Sharma, S., Sharma, S., & Athaiya, A. Activation functions in neural networks. Towards Data Sci. 2017; https://doi.org/10.33564/IJEAST.2020.v04i12.054
14. [14] Liu, C., & Hui, L. Relu soothes the ntk condition number and accelerates optimization for wide neural networks. arXiv preprint arXiv:2305.08813. 2023; https://doi.org/10.48550/arXiv.2305.08813
15. [15] He, K., Zhang, X., Ren, S., & Sun, J. Delving deep into rectifiers: Surpassing human-level performance on ImageNet classification, Proceedings of the IEEE International Conference on Computer Vision (ICCV). 2015; https://doi.org/10.1109/ICCV.2015.123.
16. [16] Kingma, D. P., & Ba, J. L. Adam: A method for stochastic optimization, International Conference on Learning Representations (ICLR). 2015; https://doi.org/10.48550/arXiv.1412.6980.
17. [17] Wang, Q., Ma, Y., Zhao, K., & Tian, Y. A comprehensive survey of loss functions in machine learning. Annals of Data Science. 2020; https://doi.org/10.1007/s40745-020-00253-5
18. [18] Erbani, J., Portier, P. É., Egyed-Zsigmond, E., & Nurbakova, D. Confusion Matrices: A Unified Theory. IEEE Access. 2024; https://doi.org/10.1109/ACCESS.2024.3507199
19. [19] Stehman, S. V. Selecting and interpreting measures of thematic classification accuracy. Remote sensing of Environment. 1997; https://doi.org/10.1016/S0034-4257(97)00083-7
20. [20] Fawcett, T. An introduction to ROC analysis. Pattern recognition letters. 2006; https://doi.org/10.1016/j.patrec.2005.10.010
21. [21] Al-Khowarizmi, Sari, I. P., & Maulana, H. Optimization of support vector machine with cubic kernel function to detect cyberbullying in social networks. Telkomnika. 2024; https://doi.org/10.12928/TELKOMNIKA.v22i2.25437
22. [22] Tieck, J. C. V., Weber, S., Stewart, T. C., Kaiser, J., Roennau, A., & Dillmann, R. A spiking network classifies human sEMG signals and triggers finger reflexes on a robotic hand. Robotics and Autonomous Systems. 2020; https://doi.org/10.1016/j.robot.2020.103566
23. [23] Li, X., Chen, S., Zhang, H., Samuel, O. W., Wang, H., Fang, P., ... & Li, G. Towards reducing the impacts of unwanted movements on the identification of motion intentions. Journal of Electromyography and Kinesiology. 2016; https://doi.org/10.1016/j.jelekin.2016.03.005
24. [24] Nasr, A., Hashemi, A., & McPhee, J. Scalable musculoskeletal model for dynamic simulations of upper body movement. Computer methods in biomechanics and biomedical engineering. 2024; https://doi.org/10.1080/10255842.2023.2184747.
25. [25] Kaya, E., & Argunsah, H. Exploring the contribution of joint angles and sEMG signals on joint torque prediction accuracy using LSTM-based deep learning techniques. Computer Methods in Biomechanics and Biomedical Engineering. 2024; https://doi.org/10.1080/10255842.2024.2400318.
26. [26] Anam, K., Swasono, D. I., Triono, A., Muttaqin, A. Z., & Hanggara, F. S. Random forest-based simultaneous and proportional myoelectric control system for finger movements. Computer Methods in Biomechanics and Biomedical Engineering. 2023; https://doi.org/10.1080/10255842.2023.2165068.
27. [27] Yıldırım, M. Y., & Akay, R., Robot Tutucu Problemi için Çok Stratejili Aritmetik Optimizasyon Algoritması. Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım Ve Teknoloji, 2024; https://doi.org/10.29109/gujsc.1383797
28. [28] Mastinu, E., Doguet, P., Botquin, Y., Håkansson, B., & Ortiz-Catalan, M., Embedded system for prosthetic control using implanted neuromuscular interfaces accessed via an osseointegrated implant. IEEE transactions on biomedical circuits and systems, 2017; https://doi.org/10.1109/TBCAS.2017.2694710.
29. [29] Nia, N. G., Kaplanoglu, E., Nasab, A., & Akgun, G., Enhancing Prosthetic Hand Control: A Study on IMU Sensor-Based Machine Learning for Precise Hand Orientation Classification, IEEE. 2024; https://doi.org/10.1109/SoutheastCon52093.2024.10500296.
30. [30] Tokunaga, D., Nishikawa, S., & Kiguchi, K., Estimating Deficient Muscle Activity Using LSTM With Integrated Damping Neurons for EMG-Based Control of Robotic Prosthetic Fingers. IEEE Access, 2023; https://doi.org/10.1109/ACCESS.2023.3312575.
31. [31] Schabron, B., Alashqar, Z., Fuhrman, N., Jibbe, K., & Desai, J., Artificial neural network to detect human hand gestures for a robotic arm control. In 2019 41st Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), IEEE. 2019; https://doi.org/10.1109/EMBC.2019.8857264.
32. [32] Sierotowicz, M., Lotti, N., Nell, L., Missiroli, F., Alicea, R., Zhang, X., ... & Masia, L., EMG-driven machine learning control of a soft glove for grasping assistance and rehabilitation. IEEE Robotics and Automation Letters, 7(2), 1566-1573. 2022; https://doi.org/10.1109/LRA.2021.3140055.
33. [33] Sree, K. S., Bikku, T., Mounika, S., Ravinder, N., Kumar, M. L., & Prasad, C. EMG controlled bionic robotic arm using artificial intelligence and machine learning. In 2021 fifth international conference on I-SMAC (IoT in social, mobile, analytics and cloud)(I-SMAC). 2021; https://doi.org/10.1109/I-SMAC52330.2021.9640623.
34. [34] Liarokapis, M. V., Artemiadis, P. K., Katsiaris, P. T., Kyriakopoulos, K. J., & Manolakos, E. S. (2012, May). Learning human reach-to-grasp strategies: Towards EMG-based control of robotic arm-hand systems. In 2012 IEEE International Conference on Robotics and Automation, IEEE. 2012; https://doi.org/10.1109/ICRA.2012.6225047.
35. [35] Li, C., Ren, J., Huang, H., Wang, B., Zhu, Y., & Hu, H. PCA and deep learning based myoelectric grasping control of a prosthetic hand. Biomedical Engineering Online. 2018; https://doi.org/10.1186/s12938-018-0539-8
36. [36] Parajuli, N., Sreenivasan, N., Bifulco, P., Cesarelli, M., Savino, S., Niola, V., Esposito, D., Hamilton, T. J., Naik, G. R., Gunawardana, U., & Gargiulo, G. D. Real-Time EMG Based Pattern Recognition Control for Hand Prostheses: A Review on Existing Methods. Challenges and Future Implementation. Sensors. 2019; https://doi.org/10.3390/s19204596