Evaluating online elasticity estimation of soft objects using standard robot grippers

• DOI: 10.48550/arxiv.2401.08298
• Publication Date: 2024-01-01
• Authors: Shubhan P. Patni; Pavel Stoudek; Hynek Chlup; Matej Hoffmann
• Keywords: Grippers; Elasticity; Tactile sensor

We experimentally evaluated the accuracy with which material properties can be estimated through object compression by two standard parallel jaw grippers and a force/torque sensor mounted at robot wrist, professional biaxial device used as reference. Gripper effort versus position curves were obtained transformed into stress/strain curves. The modulus of elasticity was different strain points effect multiple cycles (precycling), speed, gripper surface area on estimation studied. Viscoelasticity using energy absorbed in compression/decompression cycle, Kelvin-Voigt, Hunt-Crossley models. found that: (1) slower speeds improved estimation, while precycling or did not; (2) grippers, even after calibration, to have limited capability delivering accurate estimates absolute values Young's viscoelasticity; (3) relative ordering characteristics largely consistent across grippers; (4) despite nonlinear deformable objects, fitting linear approximations led more stable results than local modulus; (5) model worked best estimate viscoelasticity, from single compression. A two-dimensional space formed viscoelasticity grasp is advantageous for discrimination properties. demonstrated applicability our findings mock stream recycling scenario, where plastic, paper, metal objects correctly separated grasp, when compressed locations object. data code are publicly available.