Methods for intelligent localization and mapping during haptic exploration

Abstract

This paper presents a set of algorithms for use in simultaneous localization and mapping during haptic exploration. Several solutions are provided for the problem of a single spherical robot finger exploring a known smooth surface, starting with an unknown pose. Using pose estimates, pattern matching is performed between the robot's internal model of the surface and the known model. The robot finger is guided to explore regions of the surface that will maximize the probability of recognition. Simulation results demonstrate the effectiveness of one algorithm. In addition, it is shown that haptic exploration and dexterous manipulation can be achieved concurrently when multiple robot fingers are used.