Within the past decade, extensive research has been done on large-scale tactile sensing, as a result of which, a large variety of robot skins have been developed. These robot skins are different in various aspects: the sensing modality, interconnectivity of the sensors, modularity, the communication network, etc. This variety limits portability of software among these robot skins. In this article, a middleware is proposed that is capable of interacting in principle with any robot skin, through the use of simple drivers. Primarily, the middleware acquires data in real-time and provides its applications with those data in an abstract structure. As a result, the portability of algorithms implemented for large-scale tactile data processing is greatly increased among various available and future robot skins.

This article discusses a real-time networking infrastructure for a large-scale tactile sensing system to be used with humanoid robots. In such a system, real-time networking issues are of the utmost importance. Stemming from previous work, a theoretical model is presented and experimentally validated. Tests show real-time performance in a network of distributed computational nodes, each one in charge of managing part of the tactile system.