Journal: Frontiers in Robotics and AI

Article Title: A hybrid elastic-hyperelastic approach for simulating soft tactile sensors

doi: 10.3389/frobt.2025.1639524

Figure Lengend Snippet: General view of the robot during a grip test. It is accompanied by comparing deformation maps between a hyperelastic model (Abaqus) and an elastic model (Isaac Gym). The cylinder has protrusions of slightly varying heights, which can occasionally cause the elastic model to miss certain details (here it is shown missing a cylindrical protrusion).

Article Snippet: Empirical results from our experiments indicate that for objects with detailed geometry, the NPt value is at least 10 times the CHt value and the NPo value is at least 8 times the CHo value—necessitating a hyperelastic simulation with Abaqus, since the elastic model is generally not able to accurately simulate these fine details. illustrates the relationship between two distinct examples.

Techniques:

Journal: Frontiers in Robotics and AI

Article Title: A hybrid elastic-hyperelastic approach for simulating soft tactile sensors

doi: 10.3389/frobt.2025.1639524

Figure Lengend Snippet: An overview of the proposed pipeline. The sensors are simulated and coupled to the gripper and the robot, and subsequently, the selection criteria dictates whether to use an elastic or hyperelastic model. These models generate deformation maps, critical for producing the final synthetic tactile maps.

Article Snippet: Empirical results from our experiments indicate that for objects with detailed geometry, the NPt value is at least 10 times the CHt value and the NPo value is at least 8 times the CHo value—necessitating a hyperelastic simulation with Abaqus, since the elastic model is generally not able to accurately simulate these fine details. illustrates the relationship between two distinct examples.

Techniques: Selection

Journal: Frontiers in Robotics and AI

Article Title: A hybrid elastic-hyperelastic approach for simulating soft tactile sensors

doi: 10.3389/frobt.2025.1639524

Figure Lengend Snippet: Comparison of synthetic tactile maps from simulations with elastic and hyperelastic models for six basic objects. (a) Real 3D printed and simulated objects. (b) Deformed dielectric surface in the simulation of the object and robot in Isaac Gym. (c) Deformed dielectric surface in the simulation of the object and robot in Abaqus using a hyperelastic model. (d) Synthetic tactile map obtained from Isaac Gym using an elastic model. (e) Synthetic tactile map obtained from Abaqus using a hyperelastic model. (f) Real tactile map obtained from the actual sensor. For better visualization, the tactile maps were rotated 90° counterclockwise from the sensor’s original vertical position, as shown in . The synthetic tactile maps of rows (d,e) have been highlighted with a red border, which would be the tactile maps generated at the end of the pipeline considering the selection criteria step.

Article Snippet: Empirical results from our experiments indicate that for objects with detailed geometry, the NPt value is at least 10 times the CHt value and the NPo value is at least 8 times the CHo value—necessitating a hyperelastic simulation with Abaqus, since the elastic model is generally not able to accurately simulate these fine details. illustrates the relationship between two distinct examples.

Techniques: Comparison, Generated, Selection