QoS Metrics-in-the-Loop for Better Robot Navigation.

Navigation is an essential capability in most robotic solutions, being basic in scenarios such as inspection and maintenance, or factory intralogistics. Moreover, with the advance of service robotics and its increasing use in all kinds of real-world applications, nowadays, robots are expected to operate well in unstructured and dynamic environments, where for instance they could work close to human counterparts. In this context, traditional navigation approaches are no longer valid due to their constrained flexibility. Dealing appropriately with the intrinsic variability of open-ended environments requires robots to adapt themselves according to the situation they perceive in order to achieve the required quality of service. This paper describes a model-based framework for dealing with adaptive robot navigation. Our proposal relies on the systematic use of models for dynamically reconfiguring the robot navigation behavior, defined in terms of Behavior Trees, according to the runtime prediction and estimation of quality of service metrics based on system-level non-functional properties.