Agent programming via planning programs
Giuseppe De Giacomo, Fabio Patrizi, Sebastian Sardiña · 2010
We imagine agent 'planning' programs as programs built from achievement and maintenance goals. Their executions require the ability to meet such goals while respecting the programs' control flow. The question then is: can we always guarantee the execution of such programs? In this paper, we define this novel planningprogramming problem formally, and propose a sound, complete and optimal wrt computational complexity technique to actually generate a solution by appealing to recent results in LTL-based synthesis of reactive systems.