Image recognition of round and square objects for different grasping styles by three-fingered robot hand / Muhammad Hafiz Roshdi

This paper presents image recognition of several object shapes for grasping tasks by a three-fingered robot hand. The objective of this works is to recognize and count the number of round objects and to use the result to select the reference input position for moving the robot fingers in two different grasping styles. Image recognition on static images is implemented through programming in Simulink and M-file MATLAB. Meanwhile, Simulink program is used to send instruction to the robot main program to select the appropriate grasping style. The experimental results show that the image recognition has successfully recognized and counted the desired objects and used to produce correct output to the robot‟s main program to select for different grasping styles.