Search Results for gradient-descent-method

Abstract

This study was devoted in investigating the optimum geometric parameters for underactuated linkage three phalanges robotic finger. New kinematic and kinetic equations of grasping were derived in this research taking into account the angle for the ternary solid links of the four-bar linkages. To obtain the target of optimization, a gradient descent method was used which consists of three stages to find the optimal geometric parameters with high accuracy. Five criteria were selected to find the optimal solution by using multi objectives function algorithm, these are percentage of the grasping stability, the grasp forces, squeezing force, Mimic function for grasping task, and transmission angle for grasping operation. Gradient descent method starts by detecting the optimal geometric parameters for each criterion and choosing the best geometric parameters from the five criteria functions. At the optimum solution, the underactuated robotic finger prototype was built from hard Polylactic acid (PLA) plastic using rapid prototyping and was tested performance by grasping objects. Finally, the results have been shown that the robotic finger adapts to the wanted configurations.