Deep learning-based approach for gesture recognition with static hand representation

In the era of Artificial Intelligence (AI), our science and technology have reached to a lot of milestones, especially in the field of human-robot interaction (HRI). By fusing with the image processing techniques, AI-based strategy to enhance mutual recognition of HRI via hand signs is proposed in this investigation. Primarily, robotic hardware, theoretical computation of gripper design and vision-based techniques are introduced to establish the working environment. Then, the proposed framework including controller design and interactive platform is demonstrated. Several hand signs from human operator are collected and trained. Our approach is experimented in two cases for validating the effectiveness and properness of the proposed method with varying light condition. From these results, it can be seen obviously that this scheme is applicable in different fields such as human-aware collaboration, cognitive robot or sign language translation system.