Deep Learning-Based Traffic Management System Using Gesture Recognition

Human-robot interface technologies have advanced rapidly, enabling intelligent, efficient, and flexible solutions in many industries. These technologies include hand gesture recognition (HGR) using surface electromyography (sEMG) signals for seamless human-machine communication. For real-time applications like traffic management, prosthetic device control, and smart industrial operations, sEMG-based gesture recognition systems are more reliable, faster, and resilient to environmental variables like lighting and visual obstructions than vision-based systems. This research proposes a deep learning-based gesture detection framework for traffic management using a CNN model that optimises performance and system complexity. Five hand motions were classified by the CNN model using only four EMG channels after collecting, pre-processing, and training it with sEMG data. Its channel reduction method reduces hardware, computational load, and mobility, making it ideal for embedded and wearable applications.