Deep Tactile experience for Manipulation
Tactile sensing is inherently contact based. To use tactile data robots need to make contact with the surface of an object. In order to manipulate the object, often, robots need to make multiple contacts with the object to determine a suitable contact surface. This is inefficient and may cause damage to the object. To reduce the number of direct contacts with the object, we propose a method that, based on past experience, estimates the output of a tactile sensor from the depth data of the surface of the object.
In this project, we presented a method that estimates the output of a tactile sensor from depth sensor data. The novelty of this work lies in the way we use depth sensor data to estimate tactile sensor output. We presented qualitative and quantitative results that suggest the proposed method can estimate tactile sensor output from the depth data. We also presented results of a study that suggests reduction of point cloud density has negligible adverse affect on the quality of the tactile sensor estimates.