Design of Humanoid Hand with Adaptive Gripping Patterns Using Vision Sensor

In the realm of robotics, the emulation of human-like dexterity and adaptability remains a formidable challenge. This study presents the design of a 5-degree-of-freedom (5-DOF) humanoid hand capable of adaptive gripping through the integration of learning-based techniques. The system utilizes a camera sensor for real-time object detection, guiding the hand's gripping actions via a learning-driven mapping function. A learning-based technique is implemented for object recognition, enabling the hand to adapt to various dynamic scenarios. Experimental results demonstrate the effectiveness of the camera sensor in driving adaptive gripping actions. The proposed approach highlights the potential for more sophisticated and capable robotic systems, contributing to the evolving landscape of human-robot collaboration.