Enhancing Edge robotics through the use of context information

Abstract

Cloud robotics aims at endowing robot systems with powerful capabilities by leveraging the computing resources available in theCloud. To that end, the Cloud infrastructure consolidates servicesand information among the robots, enabling a degree of centralization which has the potential to improve operations. Despite beingvery promising, Cloud robotics presents two critical issues: (i) it isvery hard to control the network between the robots and the Cloud(e.g., long delays, high jitter), and (ii) local context information (e.g.,on the access network) is not available in the Cloud. This makeshard to achieve deterministic performance for robotics applications.Over the last few years, Edge computing has emerged as a trend toprovide services and computing capabilities directly in the accessnetwork. This is so because of the additional benefits enabled byEdge computing: (i) it is easier to control the network end-to-end,and (ii) local context information (e.g., about the wireless channel) can be made available for use by applications. The goal of this paperis to showcase, by means of real-life experimentation, the benefits ofresiding at the Edge for robotics applications, due to the possibilityof consuming context information locally available. In our experimentation, an application running in the Edge controls over a Wi-Filink the movement of a robot. Information related to the wirelesschannel is made available via a service at the Edge, which is thenconsumed by the application.Results show that a smoother drivingof the robot can be achieved when wireless quality information isconsidered as input of the movement control algorithm.