SkinSim: A simulation environment for multimodal robot skin

Abstract

Robotic skin with embedded pressure, temperature, and distance sensors are increasingly useful in safe and intuitive human-robot collaboration scenarios. Design problems for distributed arrays of skin sensors. Such as placement, optimizing resolution, networking, and use in feedback loops are difficult to solve, and highly dependent on the application. In this paper we describe SkinSim, a new simulator framework for multi-modal, multi-resolution robot skin, aimed at solving complex design problems. SkinSim is implemented using the ROS and Gazebo simulation infrastructures which are supported by the Open Source Robotics Foundation, and therefore will be shared with the community. Here we describe the overall architecture of SkinSim and its use in sensor fusion and feedback control during physical human-robot interaction. In a case study, the response of a physical tactile sensor was experimentally characterized and the corresponding reduced-order models were implemented in the simulation environment.