Compliant multi-layer tactile sensing for enhanced identification of human touch

Abstract

Tactile sensing is of interest for facilitating interactions between robots and humans. To aid the robot’s interpretation of human contact, the use of a multi-layer cutaneous tactile sensing architecture that can provide more information and an expanded force sensing range was explored, revealing differences in the signal generated by a machine versus a human. The multi-layer system consisted of two stretchable sensing skins alternating with two foam layers of different stiffness. When human touch was compared with machine indentation, a large variability in human touch was found. Thus, although the topmost skin, placed over a soft foam, was able to better recognize light contacts, a second underlying skin, placed over a harder foam, was required to gauge stronger human contacts. Out-of-plane touch modalities, such as tapping and punching, could be identified using strip-shaped skins having just two electrodes. To provide distributed sensing, the technique of electrical impedance tomography was employed with the multi-layer architecture using larger-area skins having electrodes around the perimeter. Distributed touch modalities, such as multi-point finger presses and sliding, were distinguished from single-point pressing. The distributed multi-layer sensing system also had the ability to assess higher-force touches.

Selection of optimal hyperparameter for detecting multiple contacts from large-area tactile sensors based on electrical impedance tomography

“…Various compliant functional materials have been employed to realize continuous largearea tactile sensors, including conductive fabrics [12], ionic liquids [13,14], and conductive polymer composites [15]. In addition, identification of various touch modalities [16,17] and multi-directional stimuli [18,19] from the EIT-based sensors has been demonstrated.…”

Piezoresistive Performance of a Compliant Scalable Sensor Made of Low‐Cost Exfoliated Graphite Polymer Composites

“…Currently, various functional materials have been investigated, including conductive liquids, [8] conductive thin films, [9] and conductive polymer composites. [10,11] Among them, conductive polymer composites have received increasing attention due to the advantages of both mechanical compliance and superior piezoresistivity. In addition, conductive polymer composites could be prepared by various facile fabrication techniques, [12] such as direct ink writing and screen printing, making it possible to retrofit over the surface of existing structures.…”

EIT for tactile sensing: considering artefacts in hyperparameter selection

“…Even small amounts of noise can be surprisingly detrimental to EIT images, but at what point does it become so severe that the image is no longer useful for correctly detecting the 'target' stimulus, or spatial pattern of induced impedance changes? For tactile sensing [1][2][3][4][5][6][7][8][9][10][11][12], as reviewed in [13,14], the goal of EIT is generally to provide 'actionable intelligence' from the sensing system. This may be the interpretation of a human touch or knowledge about physical contacts made by a robot with its environment.…”

“…The signal may be relatively small, arising, for example, from a human finger gently pressing on a sensor, or relatively large, as from a robot with a sensing skin bumping into an object. Furthermore, the reconstruction of each image must be sufficiently reliable for real-time inferences and decision-making, for example to understand gestures [2,5,12], so averaging many images is infeasible.…”