Tutorial on Embodiment
4.1.2. Slippage detection through skin morphology*
We have seen in previous case studies that morphology and material properties can take over a significant part of a grasping task. Nevertheless, for fine object manipulation, "blind" feed-forward control needs to be complemented with sensory feedback. We have picked slippage sensing for our case study - a prerequisite for stable grasping and fine object manipulation - and we will show how the particular shape and material properties of an artificial skin can facilitate perception.
In humans, the ridged skin structure not only improves the mechanics of grasping as mentioned above, but also magnifies the pressure (which can be perceived) exerted by the manipulated object (Fearing & Hollerbach, 1984), and acts as a frequency filter for specific skin mechanoreceptors (Scheibert et al., 2009). Similar properties are desirable in robotic or prosthetic hands. A wide range of tactile sensors have been developed for slippage detection which use different transduction principles: piezoelectric sensors sensitive to vibrations, skin with round ridges and strain sensors, vibrating nibs on the skin surface sensed by accelerometers, or brushes on top of capacitive membranes (see the references in Damian et al., 2010). The morphology and material properties are significantly involved in all of those designs. In what follows, we want to look in detail into yet another solution where morphology maximizes the information that can be acquired about a slippage event.
Damian et al. (2010) devised a tactile sensor consisting of a silicone skin layer with ridges a few millimeters apart which transduces surface events to a force sensing resistor beneath (Fig. 126.96.36.199., A). Whereas a flat skin without ridges, which was used as a reference, fails to detect an object sliding over it, ridged skin gives rise to peaks in the pressure sensor readings. Moreover, the frequency of the pressure signal obtained is directly proportional to the slippage speed and inversely proportional to the distance between ridges. The inter-ridge distance itself was found to further influence the quality of frequency encoded information. Among all skins, the one with a 4 mm spacing between ridges yielded discriminatory peak frequencies for each velocity (Fig. 188.8.131.52, C). The skin was afterwards employed in a robotic hand to stabilize grip. In summary, in this study, much of the electronic and algorithmic complexity present in other tactile sensing approaches has been successfully off-loaded to the morphology and allowed to detect slippage and gauge its speed with theoretically a single force sensor.
Fig. 184.108.40.206. Slippage detection through ridged skin. (A) Schematics of the artificial skin. Silicone skin with evenly spaced ridges is glued over a Force Sensing Resistor (FSR). (B) Robotic hand equipped with artificial ridged skin. (C) Signal generated by an object sliding over a skin without ridges (left), and with ridges 4 mm apart (right). The ridged skin provides a stronger signal with higher amplitude. In addition a clear periodic pattern allows for detection of slippage speed. (Damian et al., 2010)
* This section has been adapted from Hoffmann and Pfeifer (2011).
Damian, D.; Martinez, H.; Dermitzakis, K.; Hernandez Arieta, A. & Pfeifer, R. (2010), Artificial ridged skin for slippage speed detection in prosthetic hand applications, in 'Proc. IEEE/RSJ Int. Conf. Intelligent Robots and Systems (IROS)'.
Fearing, R. & Hollerbach, J. (1984), 'Basic solid mechanics for tactile sensing', Int. J. Robotics Research 1, 266--275.
Hoffmann, M. & Pfeifer, R. (2011), The implications of embodiment for behavior and cognition: animal and robotic case studies, in W. Tschacher & C. Bergomi, ed., 'The Implications of Embodiment: Cognition and Communication', Exeter: Imprint Academic, pp. 31-58.
Scheibert, J.; Leurent, S.; Prevost, A. & Debregas, G. (2009), 'The role of fingerprints in the coding of tactile information probed with a biomimetic sensor', Science 13, 1503-1506.