Normalizing abstractions of heterogeneous robotic systems by using Roles: usability study in the administration of software and development tools

Abstract

The modern approach for developing software in service robotics is through the use of robotics middleware platforms. Although the main concept is similar across platforms, their discrepancy among abstractions hinders the composition and administration of highly integrated systems, resulting in usability issues for the developers. We identify that this can be improved by normalizing the abstractions of the heterogeneous systems at runtime. We propose a framework that sets the concept of "Roles" to define a novel method of creating and reasoning system models by normalizing abstractions to produce practical platform-agnostic representations of systems, which is implemented in a cross-platform infrastructure and a GUI. This paper verifies and validates the functionality and usability (by benchmarking) of the proposed framework through a case-study using a real and a simulated mobile robot. The results from usage trials with test subjects showed improved usability, and demonstrated the overall advantage of using the proposed approach.