<![CDATA[Gesture recognition has emerged as a promising approach to enhance human-machine interaction, especially in robotics and assistive devices. This study presents a real-time gesture-controlled robotic system that combines deep learning and machine learning techniques to recognize hand gestures and map them to servo motor movements. A convolutional neural network (CNN) was used to classify six predefined hand gestures: a closed fist and five individual finger extensions. To enhance classification accuracy and generalization, CNNs are tuned using hyperparameter tuning techniques, such as Particle Swarm Optimization (PSO), Grey Wolf Optimization (GWO), and Brown Bear Optimization (BBO). These methods efficiently explore the hyperparameter space—such as learning rate, filter size, and batch size—reducing manual trial and error in control. Among these proposed models tested, the BBO-CNN has been achieved the highest performance with a classification accuracy of 99.98%, outperforming both PSO-CNN (99.89%) and GWO-CNN (99.44%). The model CNN without optimization achieved an accuracy of 97.50%. The combination of advanced deep learning models and embedded control demonstrates the feasibility and effectiveness of gesture-based robotics applications.]]>
Copyrights © 2025
