Component based decision architecture for reliable autonomous systems

Several decision making algorithms that are developed in robotics domain are found to be useful in automotive industrial applications. In order to use these algorithms in safety critical embedded systems one has to ensure a minimum confidence level, quality assurance, and reliability. In addition, decision algorithms are designed and developed by domain experts and system integrators do not have control over the performance of low level components. Hence, there is a compelling need for a scalable framework, which accelerates the system integration and at the same time is conform to safety levels. This position paper presents a Quality of Service (QoS) based component architecture so as to address this problem. Our approach introduces a component model that segregates the functional and non-functional aspects of decision making.