In-hand object localization: Simple vs. complex tactile sensors

How good is a tactile sensor for tactile exploration and in-hand object localization? In this paper we propose two methods that address this question. The first approach evaluates the usefulness of tactile data from different sensors by means of a Particle Filter Localization algorithm. The second approach, using an Iterative Closed Point algorithm, evaluates the synergy of tactile sensors and the geometry of robotic hands.We prove, using both approaches as a novelty, that the sensor complexity does not affect the localization performance drastically. In particular, we demonstrate that it is possible to obtain similar performance with a simple fingertip sensor, designed by us, than with more complex tactile sensors. A comparison, based on simulated data for an ideal sensor and experimental data for another three sensors, corroborates our proposal, showing the potential of our fingertip force sensor for tactile exploration and in-hand localization.