Self-reconfigurable robotics architecture utilising fuzzy and deliberative reasoning

This paper proposes a generic approach towards combining fuzzy logic and ontology-based deliberative reasoning to enable self-reconfigurability within a distributed system architecture. An Ontology-based rational agent uses outputs from a fuzzy inference system (reconfiguration layer) which passively monitors the performance of the lower-level sub-systems (application layer) in order to perform system reconfiguration. A reconfiguration is required to guarantee optimal performance within a complex robotics architecture when anomalous system and environmental changes take place. More importantly, this process of reconfiguration offers greater fault tolerance and reliability in novel scenarios as compared to isolated engineered systems. The current research work will apply the proposed framework to a visual navigation system for autonomous planetary rover missions. This demonstrates the method's success through an increase in system performance following a reconfiguration routine carried out within the application layer between two different types of visual navigation methods. Experimental analysis is carried out using real-world data, concluding that the proposed reconfigurable architecture gives superior performance against standard engineered techniques.