Reliability of uniprocessor and multiprocessor real-time artificial intelligence planning systems

By real-time artificial intelligence (AI) planning systems, we mean those systems embedded in process-control systems that must plan and execute control strategies in response to external events within a real-time constraint. We propose a methodology for estimating the reliability of uniprocessor and multiprocessor real-time AI planning systems. We first discuss why there are intrinsic faults in AI planning programs that must be considered in the reliability modeling of real-time AI planning systems. Then, we show that for uniprocessor systems, no single planning algorithm can avoid all types of intrinsic faults. Finally, we investigate a multiprocessor architecture with parallel planning with the objective of reducing intrinsic faults of real-time AI planning systems and improving the reliability of embedded systems. A robot path-planning system in static domains is used as an example to illustrate our methodology