PROTOTIPE GRIPPER UNTUK MENGGENGGAM BENDA DENGAN BERBAGAI TINGKAT KEKERASAN

• Main Author: Muliady
• Format: Theses
• Language: Indonesia
• Subjects: Teknik (Rekayasa, enjinering dan kegiatan berkaitan)
• Online Access: https://digilib.itb.ac.id/gdl/view/4964
• Institution: Institut Teknologi Bandung

<b>Abstract :</b><p align="justify"> <br /> <br /> One of the challenging problems encountered in control engineering is the problem of modeling in the case of pick and place the solid matters with various level of hardness which is commonly experienced in human knowledge qualitative and thinking process.<p align="justify"> <br /> <br /> <br /> This thesis attempts to solve the problem by using the well-known Adaptive Network-based Fuzzy Inference System (ANFIS) control to imitating human ability in grasping soft and hard materials, Using the conventional mathematic modeling in attacking the problem will complicate the system. The material that can be grasp is a solid homogeneous material which is limited by certain shape, hardness level, size and mass. The hardness characteristics of the material will be determined from the mass and the surface response of the material to the given pressure.<p align="justify"> <br /> <br /> <br /> A prototype of gripper is constructed to perfom the command created by the ANFIS controller, so the performance of ANFIS controller can be observed. The gripper system uses two strain gauge sensors and it is moved by two stepper motors for picking and grasping materials.<p align="justify"> <br /> <br /> <br /> The controller that uses the fuzzy logic is known as Fuzzy Inference System (FIS). The FIS do the mapping of input at a membership function and then carry on the fuzzy logic operations. The ANFIS will adjust the membership function using training rule of hybrid from gradient descent and least squares estimation methods, which are based on the given training data. The transfer function curve which is mapping between inputs and the gripper pressure for a successful grasping is determined by ANFIS programing.<p align="justify"> <br /> <br /> <br /> The result of the research is a prototype of gripper, which is controlled by the personal computer through an interface and an analog to digital converter. Controller software uses Simulink and Fuzzy Logic Toolbox from Matlab and communication software between hardware uses modified Real-Time Workshop from Matlab. The overall test for the system shows the level of successful in grasping 185 gram cement is 100%, 145 gram tofu is 80% and 100 gram sea weed is 20%.