An Enhanced Object Detection Approach for Robot-Hand Implementation

Recently, robot-hand has drawn many at-tention in the field of automated control and applied in different fields of industry, life, and science. Developing an automatic robot-hand with high accuracy and low cost to support people with hand disability is the interest of this paper. To this end, we introduced an enhanced image classification approach to classify targeted objects and build our own 3D Printed Robot-Hand. We are then training the robot-hand with images of typical objects and applying the output result of detecting to control the grabbing and holding processes of the robot-hand. The accuracy of object detection achieves about 91 % and the holding process is almost successful for the detected objects, which indicates the application potential of this approach.