Prosthetic Hand Based on Human Hand Anatomy Controlled by Surface Electromyography and Artificial Neural Network
[EN] Humans have a complex way of expressing their intuitive intentions in real gestures. That is why many gesture detection and recognition techniques have been studied and developed. There are many methods of human hand signal reading, such as those using electroencephalography, electrocorticograp...
| Authors: | , , , |
|---|---|
| Format: | article |
| Publication Date: | 2025 |
| Country: | España |
| Institution: | Universitat Politècnica de València (UPV) |
| Repository: | RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia |
| Language: | English |
| OAI Identifier: | oai:riunet.upv.es:10251/230278 |
| Online Access: | https://riunet.upv.es/handle/10251/230278 |
| Access Level: | Open access |
| Keyword: | Prosthetic hand Bionic hand EMG Gesture recognition Feature extraction Classification 03.- Garantizar una vida saludable y promover el bienestar para todos y todas en todas las edades 09.- Desarrollar infraestructuras resilientes, promover la industrialización inclusiva y sostenible, y fomentar la innovación |
| Summary: | [EN] Humans have a complex way of expressing their intuitive intentions in real gestures. That is why many gesture detection and recognition techniques have been studied and developed. There are many methods of human hand signal reading, such as those using electroencephalography, electrocorticography, and electromyography, as well as methods for gesture recognition. In this paper, we present a method based on real-time surface electroencephalography hand-based gesture recognition using a multilayer neural network. For this purpose, the sEMG signals have been amplified, filtered and sampled; then, the data have been segmented, feature extracted and classified for each gesture. To validate the method, 100 signals for three gestures with 64 samples each signal have been recorded from 2 users with OYMotion sensors and 100 signals for three gestures from 4 users with the MyWare sensors. These signals were used for feature extraction and classification using an artificial neuronal network. The model converges after 10 sessions, achieving 98% accuracy. As a result, an algorithm was developed that aimed to recognize two specific gestures (handling a bottle and pointing with the index finger) in real time with 95% accuracy. |
|---|