On-line sensing and modeling of mechanical impedance in robotic food processing

Abstract

Measurement of mechanical impedance is useful in robotic food processing. In cutting meat, fish, and other inhomogeneous objects by means of a robotic cutter, for instance, it is useful to sense the transition regions between soft meat, hard meat, fat, shin, and bone. Product quality and yield can be improved through this, by accurately separating the desired product from the parts that should be discarded. Mechanical impedance at the cutter-object interface is known to provide the necessary information for this purpose. Unfortunately, the conventional methods of measuring mechanical impedance require sensing of both force and velocity simultaneously. Instrumenting a robotic end-effector for these measurements can make the cutter unit unsatisfactorily heavy, sluggish, and costly. An approach for on-line sensing of mechanical impedance, using the current of the cutter motor and the displacement (depth of cut) of the cutter, has been developed by us. A digital filter computes the mechanical impedance on this basis. For model-based estimation, performance evaluation of the on-line sensor, and also for model-based cutter control, it is useful to develop a model of the cutter-object interface. This paper illustrates these concepts using a laboratory system consisting of a robotic gripper and a flexible object. The prototype consists of an industrial-quality robotic gripper, a control computer, and associated hardware and software for data acquisition and processing. A model of the process interface between the end-effector and object has been developed. Some illustrative results from laboratory experiments are given.