Achieving texture recognition through processing of tactile information could significantly improve robotic prehensile and manipulative capabilities. By producing an object signature based on such information, mishandling due to friction or slippage could be avoided. However, this would require acquisition and processing of tactile data in close to real time in order to function at task speed. This paper proposes a new texture-discriminating algorithm that requires very little exploratory movement. We compared the success rate of two types of exploratory movement for the recognition textures with directional properties such as grooves. Another goal of this study was to obtain an algorithm that is largely insensitive to the velocity and contact force of the sensor movement. We used a genetic algorithm to optimize the variables and the topology of our neural network. We improved the results with a new approach to majority voting that does not require numerous samples. Object classification was more than 90 % correct and most of the errors involved textures that humans are barely able to differentiate.

Roughness estimation can help with improving tactile prehension and distinguishing slippage events during object manipulation with a robotic hand. Humans are able to estimate roughness from a small contact area with an object, and adapt manipulation strategies using this information [1]. In order to do the same with a robotic hand fitted with tactile sensors, this article focuses on how to estimate roughness with data from a tactile sensor. We propose a learning algorithm that estimates roughness on a scale from 1 to 5, which was inspired by human tactile capabilities. For more adapted parameters values, this algorithm is optimized with a genetic algorithm. To initialize the scale, we asked 30 people to classify 25 textures on a roughness scale from 1 to 5. The results were used to feed the learning algorithm. After testing our algorithm on those 25 textures, we conclude that even if there are small errors on certain textures, our algorithm is able to adapt itself to new textures and provide a roughness estimation that approximates the human one.

Robotic operations that involve the displacement of objects generate different kinds of dynamic events. These may simply correspond to normal robot-related motion, or contact(s) with the object(s) during grasping, but they may also be potentially-problematic events like slippage. In this paper, we use sparse data from tactile sensors to detect slippage and discriminate object-gripper slip from object-world slip. The method we propose can also identify vibrations that correspond to other dynamic events automatically, even when those events are not related to slippage. The tactile data can then be classified, allowing the robot to react accordingly. To achieve this goal, we compute the power spectral density (PSD) of the tactile dynamic signal, and we apply transformations to the PSD that were inspired by the automatic speech recognition (ASR) field. The originality of this work comes from using a sparse representation of the transformed data to obtain sparse vectors containing a small set of high-level features. Those sparse vectors are then used as inputs to a simple linear support vector machine (SVM), that acts as a classifier and quickly estimates the event to which they correspond. Our method was tested on data obtained from 244 experiments that were conducted on 32 different everyday-objects. Results show that we can successfully discriminate most of the dynamic events we studied in this work. Moreover, by using this technique, we are able to detect slippage with an accuracy of 92.60% and to differentiate object-gripper slip from object-world slip with a success rate of 89.42%.