Data-driven Super-resolution on a Tactile Dome
This addresses the need for improved tactile sensing in robotics, particularly for integration into hands or fingers, though it appears incremental as it builds on existing sensor technology with data-driven enhancements.
The paper tackles the problem of low spatial resolution in tactile sensors for robotic manipulation by presenting a method that localizes contact with high accuracy on curved surfaces, achieving a best-case accuracy of 1.1mm over a 1,300mm² hemisphere.
While tactile sensor technology has made great strides over the past decades, applications in robotic manipulation are limited by aspects such as blind spots, difficult integration into hands, and low spatial resolution. We present a method for localizing contact with high accuracy over curved, three dimensional surfaces, with a low wire count and reduced integration complexity. To achieve this, we build a volume of soft material embedded with individual off-the-shelf pressure sensors. Using data driven techniques, we map the raw signals from these pressure sensors to known surface locations and indentation depths. Additionally, we show that a finite element model can be used to improve the placement of the pressure sensors inside the volume and to explore the design space in simulation. We validate our approach on physically implemented tactile domes which achieve high contact localization accuracy ($1.1mm$ in the best case) over a large, curved sensing area ($1,300mm^2$ hemisphere). We believe this approach can be used to deploy tactile sensing capabilities over three dimensional surfaces such as a robotic finger or palm.