New Insights in Robot Haptic Object Discrimination!
Do you think haptic object recognition is about the sensory features or the classifier? Comparing two anthropomorphic hands and two 2-finger grippers, here we show that the gripper embodiment and the action parameters (hand configuration and grasping speed) play a far bigger role!
Citation: Pliska, M., Patni, S. P., Mares, M., Stoudek, P., Straka, Z., Stepanova, K. and Hoffmann, M. (2024), 'Single-grasp deformable object discrimination: the effect of gripper morphology, sensing modalities, and action parameters', IEEE Transactions on Robotics 40, 4414 - 4426.
Full text: [DOI - IEEE Xplore][pdf-arxiv]
Data: https://osf.io/zetg3/
In this study, using four different robot grippers (2 parallel jaw, 2 anthromorphic hands) we explored how gripper embodiment, sensory inputs, and movement parameters (grasp configuration and speed) impact the ability to discriminate objects. We conducted over 24,000 measurements on two datasets, including one consisting of 20 polyurethane foams - practically indistinguishable for humans.
💡 Visualization of the feature space showed that gripper shape and movement were the biggest contributors to variance, highlighting the challenge of generalization (Cross-Embodiment) in haptic perception.