Material comparison and design of low cost modular tactile surface sensors for industrial manipulators

Physical human robot interaction (pHRI) in industrial manufacturing environments requires reliable environment perception capabilities usually employing multiple sensor modalities and particularly tactile sensor matrices to facilitate the management of human robot contacts. However, their high cost still impedes the large-scale integration of pHRI in manufacturing environments. To enable the development of low-cost tactile sensor matrices for pHRI applications, we examine five cost-effective piezoresistive materials w.r.t. their suitability for tactile sensor matrices. Since hysteresis and drift can severely deteriorate the performance of a tactile sensor, particular attention is paid to these properties. For the apparently best material and a very low-cost alternative, the input-output behavior is modeled using a machine learning approach. The differences in the input-output behavior of the individual taxels are comparatively low - thus significantly simplifying parameter identification and calibration.