A Review of Pattern Recognition Control Methods for Activation of Instrumented Wheelchair Power Assist Systems Based on sEMG Reading

Abstract

Power-assisted wheelchairs have transformed rehabilitation technologies, providing people with disabilities more freedom and mobility. Signals from surface electromyography (sEMG) are vital for providing intuitive control over these systems. With an emphasis on classification accuracy, computational requirements, and suitability for instrumented wheelchair systems, this review assesses several pattern recognition techniques such as KNN, SVM, LDA, LSTM, Decision Trees, and Artificial Neural Networks (ANN). The study highlights the growing significance of hybrid approaches that combine pattern recognition techniques to increase robustness and precision. Even though KNN has the highest accuracy, methods such as LSTM and SVM are more effective and versatile, making them more appropriate for real-time applications. The results highlight the necessity of more research into personalized systems and hybrid models, which have huge potential to advance assistive technologies.