Gesture Prediction Using Wearable Sensing Systems with Neural Networks for Temporal Data Analysis.

Area of Science:

• Wearable electronics
• Human-computer interaction
• Artificial intelligence

Background:

• Interactive systems require reduced delays, necessitating human gesture prediction.
• Hand gesture recognition is crucial for intuitive human-computer interaction.
• Existing systems need to capture complex and rapid hand movements effectively.

Purpose of the Study:

• To demonstrate real-time hand gesture prediction using artificial neural networks (ANNs).
• To utilize data from wearable pyrolytic graphite sheet (PGS) based data gloves for gesture prediction.
• To evaluate the efficacy of ANNs in forecasting hand gestures from sensor resistance data.

Main Methods:

• Fabrication of strain sensors and wearable devices using pyrolytic graphite sheets (PGSs).
• Collection of hand gesture data from subjects using PGS-based data gloves.
• Development and implementation of a four-layered artificial neural network (ANN) for gesture prediction.

Main Results:

• The developed system successfully predicted hand gestures in real time.
• The ANNs demonstrated high sensitivity, durability, and fast response in detecting gestures.
• Comparison with other algorithms confirmed the proposed ANN method's effectiveness.

Conclusions:

• Hand gesture prediction systems based on ANNs and PGS wearable devices are feasible and effective.
• The proposed system can forecast various hand gestures using resistance data from wearable sensors.
• This technology offers a promising approach to reduce delays in interactive systems.