Design Of A Anthropomorphic Robotic Gripper With Pressure Sensor

Robotic manipulator is one of the most important components in a robotic system. It performs various material handling jobs with dexterous moves, to replace humans in adverse environments. The "perception" plays a critical role in intelligent control of robotic grippers. A variety of perceptual technologies have been developed and applied for robotic control. Pressure is one of the most basic parameters used in robotic arm control, because it is simple, reliable and quick response. Pressure, however, has not been given sufficient attention in control of grippers. In most related research works, pressure is "sensed" but not "comprehended". In general, it lacks fundamental studies and constructional technologies in this area.In this thesis pressure sensors are used in combination with Surface electromyography signal (sEMG for short) to measure, comprehend and control a robotic gripper. This approach can result in higher accuracy, resolution and controllability, quicker response, and more safety to the surroundings. Specifically this study includes:(1) A quantitative mapping is established between various items’ weights and sEMG. To use sEMG for gripper control and prediction the weight of objects, sEMG signals are collected first and the noise reduction is performed. And RMS is applied as the feature extraction method. After using sEMG threshold to control manipulator to move, eight persons are asked to grip twenty one different objects. sEMG of each person’s is measured. The mapping between sEMG and the weights of the objects is hence established. And a linear fitting function is built up as well.(2) A relationship is created between fingertip pressures and the weights of a variety of items. Following the "infinite approximation" approach, the robotic gripper is tested to grasp an object using different amount of force, until a minimal force is found that enable the gripper to collect the weight without slip. The friction coefficient is then calculated for this type of item. A variety of items are tested to figure out their coefficients. With such coefficients available, the gripper equipped with pressure sensors may be controlled to pick up those types of items in any weights with proper holding forces.(3) Using the above as the fundamental, a pressure-based perceptual system is achieved for feedback control. The robotic gripper can "feel" and "know" the objects whether they are handled or not, like a human hand would do. In addition, real-time motion control of a gripper is realized by using sEMG threshold control with pressure as feedback. A pressure sensory feedback control system is established. Intensive experiments are conducted. The results have verified the feasibility of the pressure-based perceptional control approach developed in this thesis.