Knowledge-based planning systems: an engineering perspective

Planning is an important task in many industrial settings and has been an attractive target area for the application of artificial intelligence (AI) techniques, since the early 1960s. Yet little progress has been achieved in terms of practically useful planning systems. The author reviews the initial attempts of AI researchers to develop general purpose planning systems, and explains why this approach failed to produce useful planners for engineering applications. A discussion is given on a few of `expert planning systems´ developed in the mid 1980s and early attempts to abstract expert planning shells from them. A look is taken at the architecture, representation and reasoning of four third generation AI planning systems that use the model-based reasoning (MBR) approach. The author concludes that MBR planning systems have the potential to evolve into powerful `knowledge-based planning shells´ that can interface with CAD and other computer tools used in engineering design