Runtime Analysis and Adaptation of a Hard Real-Time Robotic Control System

The increasing complexity of technical systems often leads to problems when they are to be maintained, changed or extended. Nature inspired concepts like self-management or self-organization have found their way into technical systems to overcome the complexity problems. In turn those systems exhibit beneficial self-* properties like self-optimization, self-healing or self-protection. This paper presents a software architecture for the control of parallel kinematic machines and its evolvement to a self-adaptive system that strives to optimize, protect and heal itself. A software engineering approach for the development of self-managing components is introduced that is supported by behavior validation in a specialized simulation environment. The first realization of a self-manager, responsible for the distribution of control components, is described in detail to show that self-management is feasible in robotic control. The self-manager uses formal analysis techniques during the runtime of the system to make sure it always conforms to its real-time requirements.