Alternative Sensing and Perception based on Physical fields for Robotics and Automation

Abstract

Recent advances in sensing and perception systems (SPSs), which move beyond from point measurements to field representation, enable exciting new technologies to facilitate autonomous machines capable of evolving with more and more ‘smart functions’ that ultimately make the process a self-improved system. This talk introduces several SPS methods based on physical fields as an alternative or a complement to visible light which is commonly assumed as the medium in conventional machine vision. As will be illustrated with practical robotics/automation examples, physical fields that exist both in man-made systems and in nature can be reconstructed from limited measurements to infer motion variables for guiding navigation and/or identify system properties of a distributed-parameter system to control its system behaviors; thus, their creative uses can eliminate costly, complicated external measurement systems. Selected applications of physical fields, which include electromagnetic and displacement fields in intelligent manufacturing applications and geomagnetic and thermal fields in nature, are given to help illustrate the impacts and yet to cover a wide variety of applications. The objective is to stimulate discussion of the future SPS research and its emerging applications to address problems facing society in a rapidly changing world.