Temporal-Numeric Planning with Infinite Domain Action Parameters

A limitation of task planning as a search problem is that action parameters are restricted to take their values from a finite collection of objects defined in a problem instance. In real-world, there are parameters that have freedom of taking their values from infinite domains, which we call control parameters. Planning with control parameters are deemed to be more than just a modelling choice; but it is a modelling requirement in real-world applications. POPCORN is a partially-ordered forward state space search planner that can reason with multiple typed control parameters declared in action schemes. In this paper, we briefly introduce the planning with control parameters paradigm.