Hand Gesture Identification for Soft Controller regarding Human Behaviours

In the context of the Internet of Things (IoT), gesture control offers a natural and convenient interaction method.Compared to traditional physical buttons or touchscreens, gesture control is more intuitive and flexible, making it highly suitable for smart clothing applications.This study focuses on analyzing the reliability of PET (Polyethylene terephthalate)-based piezoresistive thin-film sensors for hand gesture recognition, demonstrating their potential for smart clothing.A survey identified 13 common hand gestures frequently used in daily activities.Experimental results showed that the sensors effectively recognized these gestures, achieving a recognition accuracy of 99.4% through neural network modeling with the original design of 25 sensors.To simplify the design and improve reliability, a greedy algorithm was used to find a locally optimal solution, reducing the number of sensors from 25 to 6 while maintaining a recognition accuracy of 97%.This optimization significantly reduced system complexity, lowered costs, and made the product more environmentally friendly.In reliability testing, the original 25-sensor design had a failure rate of 7.6910 -5 , with the first failure occurring after 13,000 uses.The optimized design with 6 sensors exhibited a slightly improved failure rate of 7.6410 -5 , with the first error appearing after 13,082 uses.As testing continued, error fluctuations increased in both layouts, indicating that long-term sensor performance degrades over time.However, the optimized design notably enhanced the system's durability and reliability.In conclusion, this study confirms that PET-based sensors are not only reliable but also benefit from sensor quantity optimization, improving system durability and cost-effectiveness.These qualities make them highly suitable for integration into smart clothing's remote-control systems, offering a lightweight, durable, and efficient solution for future innovations.