Safety engineering of computational cognitive architectures within safety-critical systems

Abstract

Listen

Translate article

This paper presents the integration of a cognitive architecture with an intelligent decision support model (IDSM) that is embedded into an autonomous non-deterministic safety critical system. The IDSM will integrate multi-criteria decision making via intelligent technologies like expert systems, fuzzy logic, machine learning and genetic algorithms.Cognitive technology is currently simulated in safety–critical systems to highlight variables of interest, interface with intelligent technologies, and provide an environment that improves a system’s cognitive performance. In this study, the IDSM is being applied to an actual safety–critical system, an unmanned surface vehicle (USV) with embedded artificial intelligence (AI) software. The USV’s safety performance is being researched in a simulated and a real world nautical based environment. The objective is to build a dynamically changing model to evaluate a cognitive architecture’s ability to ensure safe performance of an intelligent safety–critical system. The IDSM does this by finding a set of key safety performance parameters that can be critiqued via safety measurements, mechanisms and methodologies. The uniqueness of this research will be on bounding the decision making associated with the cognitive architecture’s key safety parameters (KSP).Other real-time applications that could benefit from advancing the safety of cognitive technologies are unmanned platforms, transportation technologies, and service robotics. The results will provide cognitive science researchers a reference for safety engineering artificially intelligent safety–critical systems.