Bayesian Exploration for Intelligent Identification of Textures

Abstract

In order to endow robots with human-like abilities to characterize and identify objects, they must be provided with tactile sensors and intelligent algorithms to select, control, and interpret data from useful exploratory movements. Humans make informed decisions on the sequence of exploratory movements that would yield the most information for the task, depending on what the object may be and prior knowledge of what to expect from possible exploratory movements. This study is focused on texture discrimination, a subset of a much larger group of exploratory movements and percepts that humans use to discriminate, characterize, and identify objects. Using a testbed equipped with a biologically inspired tactile sensor (the BioTac), we produced sliding movements similar to those that humans make when exploring textures. Measurement of tactile vibrations and reaction forces when exploring textures were used to extract measures of textural properties inspired from psychophysical literature (traction, roughness, and fineness). Different combinations of normal force and velocity were identified to be useful for each of these three properties. A total of 117 textures were explored with these three movements to create a database of prior experience to use for identifying these same textures in future encounters. When exploring a texture, the discrimination algorithm adaptively selects the optimal movement to make and property to measure based on previous experience to differentiate the texture from a set of plausible candidates, a process we call Bayesian exploration. Performance of 99.6% in correctly discriminating pairs of similar textures was found to exceed human capabilities. Absolute classification from the entire set of 117 textures generally required a small number of well-chosen exploratory movements (median = 5) and yielded a 95.4% success rate. The method of Bayesian exploration developed and tested in this paper may generalize well to other cognitive problems.

Active Bayesian Perception for Simultaneous Object Localization and Identification

“…Although that study was neither Bayesian nor explicitly 'where' and 'what', we see their biomimetic control as analogous to the 'fixation point' strategy considered here. The other study used 'Bayesian exploration' to select between motions to best perceive surface texture with a biomimetic fingertip [4]. Their algorithm for Bayesian exploration has a similar looped architecture to that used here, but instead employs a 'disambiguation' control strategy to solve a solely 'what' task (the locations are known).…”

Active sensorimotor control for tactile exploration

“…These simulated and real-time validations demonstrate the 33 benefits of active over passive perception to successfully accomplish an autonomous exploration task using tactile 34 sensing. Furthermore, the implementation of the proposed sensorimotor control strategies indicates how the weight 35 and reliability of the information improve the perceptual accuracy and speed during the exploration task. The findings 36 of this investigation provide new robust and accurate computational methods to exploit the advances and integration 37 of tactile sensing technology in robotics.…”

Inspection System for Benchmarking of Perceived and Technical Characteristics of Surfaces

“…The BioTac thus tries to mimic different functionalities of the human mechanoreceptor system. Its developer Fishel was able to show, that it is possible to discriminate different textures with very high accuracy (>99 %) using the BioTac (requirement 4) [22]. However, the sensors accuracy coupled with a robot as the movement providing device, remains to be evaluated since the robot will probably generate significant noise in the vibration signal due to acceleration an deceleration.…”