Task-Oriented Algorithm for Two Human-Like Ambidextrous Robotic Hands Control

Abstract

Task-oriented grasp analysis was made concentrating on the reduction of complexity to achieve adaptable and stable grasps. The hand orientation normal was used for the alignment between the hand and the object. A Varying Pulse Width Form was used to represent the object, where the values of 3D model orientation angles were converted into pulse width values (VPWF) of the objects, and an object matching loop (OML) was used to choose the most suitable hand shape, using defined splines of the overall 3D model of the object, and a curvature adaptation algorithm. The experiments showed that we can analyze the gestures of grasping regardless of grasping power, and showed that there is a strong relation among finger joints. The experiments also showed that coordinating between the two hands based on the Normal Inclination method offers a more generic, speedy, and precise solution for a wide range of robotic applications. One of the important results was in deriving a relation of the precision of the grasp due to the object angle of orientation and brightness. The final aim of this research is to integrate the control of both arms and hands as one entity, avoid collision, coordinate between them, and use them in any application where labor involves a specific series of tasks: for rescuing, manufacturing, making surgery, and artificial limbs.