Single-Grasp Object Classification and Feature Extraction with Simple Robot Hands and Tactile Sensors

Abstract

Classical robotic approaches to tactile object identification often involve rigid mechanical grippers, dense sensor arrays, and exploratory procedures (EPs). Though EPs are a natural method for humans to acquire object information, evidence also exists for meaningful tactile property inference from brief, non-exploratory motions (a 'haptic glance'). In this work, we implement tactile object identification and feature extraction techniques on data acquired during a single, unplanned grasp with a simple, underactuated robot hand equipped with inexpensive barometric pressure sensors. Our methodology utilizes two cooperating schemes based on an advanced machine learning technique (random forests) and parametric methods that estimate object properties. The available data is limited to actuator positions (one per two link finger) and force sensors values (eight per finger). The schemes are able to work both independently and collaboratively, depending on the task scenario. When collaborating, the results of each method contribute to the other, improving the overall result in a synergistic fashion. Unlike prior work, the proposed approach does not require object exploration, re-grasping, grasp-release, or force modulation and works for arbitrary object start positions and orientations. Due to these factors, the technique may be integrated into practical robotic grasping scenarios without adding time or manipulation overheads.

Learning to Grasp Without Seeing

“…• Hand Crafted. This is from [29], where raw signals from three specific events (before contact, when the finger closing movement is stalled due to object-finger contact, after the fingers are in equilibrium) are extracted. Choice of Classifiers.…”

iCLAP: shape recognition by combining proprioception and touch sensing

“…Both tactile features and joint configurations are clustered by SOMs and classified by ANNs separately and the classification results are merged to achieve a final decision. In a more recent work [39], the actuator positions of robot fingers and tactile sensor values form the feature space to classify object classes using random forests but there are no exploratory motions involved, with data acquired during a single and unplanned grasp.…”

Iterative Closest Labeled Point for tactile object shape recognition

“…In [21], the proprioceptive data (finger configurations/positions) and tactile features for a whole palpation sequence were concatenated into one description for object classification; the information of tactile features and contact points is combined but the information of the positions where specific tactile features were collected was lost. In recent work [22], an underactuated robot hand, with a row of TakkTile tactile sensors embedded in each link of robot fingers, was employed for object classification. The actuator positions and force sensor values form the feature space to classify object classes using random forests but there were no exploratory motions involved, with data acquired during a single and unplanned grasp.…”

“…In total, four different features are used and compared, i.e., the Tactile-SIFT descriptors proposed in our previous work [17] and three previous features in the literature, i.e., Zernike moments (the best performing feature used in [15]), normalized Hu's moments [22], raw image moments (up to order 2) [23]. Based on [17], the dictionary size k was set to 50 through the experiments.…”

“…6. The recognition rates with tactile information only (BoW) against different number of touches, using our Tactile-SIFT descriptors [17], Zernike moments [15], normalized Hu's moments [22] and raw image moments [23] respectively.…”