Enabling Intrinsic Tactile Sensing with Soft Optical Tactile Sensors

Robotic tactile sensing can be grouped into intrinsic, placed within the mechanical structure of the robot, estimating contact locations and forces from force sensors, and extrinsic, mounted at the contact interface, and dealing with intrinsic tactile data. In the former category, it is worth mentioning Intrinsic Tactile Sensing (ITS). ITS is a technique that relies on force/torque measurements, and on a priori knowledge of the geometry of the exploring surface, to approximate the distribution of compressive tractions on the surface with a resultant contact force and moment applied to the contact centroid (contact sensing problem). These quantities are fundamental for grasping planning, control, and assessment. ITS is a well-established technique with rigid surfaces, but it is ill-suited for soft deformable materials, although some solutions to deal with this case have been proposed. For the extrinsic sensors, soft optical tactile sensors, which exploit vision information to infer contact properties, have emerged as a promising solution to estimate object tactile properties, but they have never been used to address the contact sensing problem. Integrating the characteristics of ITS with soft optical tactile sensing could significantly advance the perceptual and grasping properties of robotic hands endowed with soft fingers. This work aims to bridge this gap, proposing the integration of ITS with the TacTip, a marker-based soft optical sensor. Experiments that validate the here proposed approach are discussed.