Intelligent gripper design and application for automated part recognition and gripping

摘要

Intelligent gripping may be achieved through gripper design, automated part recognition, intelligent algorithm for control of the gripper, and on-line decision-making based on sensory data. A generic framework which integrates sensory data, part recognition, decision-making and gripper control to achieve intelligent gripping based on ABB industrial robot is constructed. The three-fingered gripper actuated by a linear servo actuator designed and developed in this project for precise speed and position control is capable of handling a large variety of objects. Generic algorithms for intelligent part recognition are developed. Edge vector representation is discussed. Object geometric features are extracted. Fuzzy logic is successfully utilized to enhance the intelligence of the system. The generic fuzzy logic algorithm, which may also find application in other fields, is presented. Model-based gripping planning algorithm which is capable of extracting object grasp features from its geometric features and reasoning out grasp model for objects with different geometry is proposed. Manipulator trajectory planning solves the problem of generating robot programs automatically. Object-oriented programming technique based on Visual C++ MFC is used to constitute the system software so as to ensure the compatibility, expandability and modular programming design. Hierarchical architecture for intelligent gripping is discussed, which partitions the robot’s functionalities into high-level (modeling, recognizing, planning and perception) layers, and low-level (sensing, interfacing and execute) layers. Individual system modules are integrated seamlessly to constitute the intelligent gripping system.