The first International Planning Competition (IPC-1) was held in 1998, organized by Drew McDermott of Yale University and run alongside the AIPS conference in Pittsburgh, the meeting series that later became ICAPS. Five planning systems competed. The competition has run roughly every two years since, and the official ICAPS record of the series is the source cited here.
The competition’s most durable product was a common language. To compare planners fairly, McDermott and colleagues created the Planning Domain Definition Language (PDDL), a standard, Lisp-like notation for describing planning problems: the objects in a world, the actions available with their preconditions and effects, the starting state, and the goal. PDDL was inspired directly by the STRIPS action representation from 1971, generalizing it so that many research groups could state problems the same way. With a shared language and a shared set of benchmark problems, planners could finally be measured head to head rather than each demonstrated on its author’s hand-picked examples.
That shift had a large effect on the field. The competition created a feedback loop that drove rapid progress in planning algorithms, especially fast heuristic search planners, and PDDL became the lingua franca of automated planning research. Successive competitions added tracks for planning under uncertainty, with numeric resources, and with hierarchical structure.
Why business readers should care: shared benchmarks and a common problem language are how a research area turns vague claims into measurable progress, the same pattern later seen in machine-learning leaderboards. IPC professionalized AI planning, the technology behind automated scheduling and logistics.