Abstract:
When a robot manipulates a flexible object, a delicate grasp is required. Slip is an important indicator that can enable a robotic manipulator to delicately grasp an obje...Show MoreMetadata
Abstract:
When a robot manipulates a flexible object, a delicate grasp is required. Slip is an important indicator that can enable a robotic manipulator to delicately grasp an object. Therefore, we created a 3×3 tactile sensor matrix covered with a elastomer grip, intended to be used on a robotic fingertip, to detect slip. We obtained tactile sensor values corresponding to the slip and nonslip states in advance and created a training dataset with several patterns. Random Forest Classifiers were trained on the datasets, and the results were compared. All nine sensor elements in the sensor matrix could detect slips with practical accuracy in the real-time robot experiments. In time periods wherein no slip occurred, no nonslip states were erroneously determined as slip states under any of the tested conditions. The slip detector created in this study was demonstrated to be applicable to types of rope that were not used to train the detector.
Date of Conference: 14-18 October 2024
Date Added to IEEE Xplore: 25 December 2024
ISBN Information: