Handwriting is one among the most fine-motor work which needs precision, stability and coordination. People with neural disorders have a difficulty in keeping their handwriting steady. This paper discusses the design and development of an assistive glove for precise handwriting. This wearable glove model uses deep learning for checking and improving fine-motor skills. The glove is integrated with two Inertial Measurement Unit (IMU) sensor, flex sensors, and pressure sensors to collect the data of wrist position, finger joint movement, and grip strength during handwriting. The collected data go through preprocessing, normalization, and feature extraction. They are then analysed through an LSTM network for learning sequences over time and classifying movements. The glove is then tested with ten users writing the English alphabets where each user has different speed, different style and different strokes. After training the model achieved an overall accuracy of 81% and a prediction accuracy of 93.7%. These results show that the glove can accurately identify and distinguish handwriting patterns. This framework shows the utilization of sensor integration and LSTM-based modelling for motor rehabilitation and analysing handwriting performance.
S et al. (Thu,) studied this question.