Variability Management in Robotic Systems: A Variability-Modelling Language That Implements Variation Points Based On Binding Time and Binding Mode

Abstract

Technological advancements have led to a growing demand for efficient solutions that minimise risk and maximise efficiency when it comes to performing tasks in nondeterministic environments. In the wake of the current pandemic, humanity has been affected by labor shortages, logistic challenges, unexpected strain on supply systems and product shortages caused by safety regulations and health complications. Undeniably, these challenges have had a resounding effect across the global service industry; hence the emergence of viable robotic solutions to alleviate these problems. Service robots are a category of robots that render services to humans. Service robots are often designed to operate in highly heterogeneous environments in collaboration with humans, or other robots. The effective completion of tasks by service robots may involve the combination of a specified set of robotic capabilities. These capabilities are mainly driven by robotic features. Valid combinations of robotic features to fit different contexts, give rise to some level of variability—i.e., the ability of a software artifact to be changed to fit different contexts, environments, or purposes. This constitutes a possible strategy to enable robotic applications to be changed, customized, or configured to fit different scenarios. Thus the need to have an effective mechanism for planning, designing, and implementing variability. Cognisant of this fact, we present a technique that implements variation points with reference to binding time and binding mode. As a long term goal, this implementation comes as an extension of the Self-adaptive dEcentralized Robotic Architecture (SERA) framework architecture. SERA, which is a decentralized architecture that supports the building of autonomous, heterogeneous, and collaborative robotic applications, lacks the ability to manage variability dynamically. For that matter, SERA and its ilk typically do not provide roboticists with the means and techniques required to manage variability effectively. With a design science approach, we define example systems as feature models, study the variability of these feature models, and then proceed to conceive a variabilitymodelling language, that provides mechanisms for managing variability with respect to binding time and mode. Our variability-modelling language is offered as an opensource library that provides basic support for binding features in robotic systems. This study provides evidence of the extensibility of robotics reference architectures to support variability in a domain where variability is typically performed in an ad hoc v manner. Its implementation is expected to alleviate extension complexity, reduce performance costs, and minimize resource consumption in robotic systems while giving roboticists the flexibility boost they so desire when it comes to engineering robotic systems. Furthermore, this conclusive study will provide evidence to back the claims that the proposed variability management technique is novel, realizable, useful in practice and provides a means for assessing configuration validity.