TASK ORIENTED TACTILE SENSING FOR A ROBOTIC GRIPPER

The sense of touch is a very useful property of the human hand that is important for grasping and manipulating a variety of objects.The adaptation of this ability in mechani­cal grippers would facilitate skillful performance of operations by robots in unstructured environments. This thesis describes the design, development and implementation of a task oriented procedure to determine the status of a robotic gripper during grasping and releasing operations. The task status obtained using this procedure, provides a complete set of information about the task including the direction of object displacement relative to the gripper. The computer aided procedure uses the dynamic forces measured during grasping and releasing operations. A prototype gripper system was designed and built to measure the real time dynamic forces during selected tasks. An integrated computer program performed tactile imaging and image interpretation to facilitate decision making based on the force data. The decisions obtained from each block of data were displayed in the form of a set of task status parameters describing grasping and releasing levels with associated confidence factors. Using the task status information,a method was designed and im­plemented to determine a set of control decisions to aid in the completion of a task. The computer aided procedure is a useful contribution to the development of intel­ligent grippers.The suitability of the procedure has been demonstrated using a set of control decisions obtained from the expert system. Since the procedure was validated using experimental data obtained while performing real tasks on a set of selected laboratory samples,its applicability in practical industrial environments is promising