Three-Dimensional Visualization of Hierarchical Task Network Plans

Artificial Intelligence (AI) planning techniques, particularly hierarchical task network (HTN) planners, can produce very complex plans, with many sequential and parallel actions decomposed into multiple levels of abstraction. Most such planners provide only a twodimensional (2D) display of such plans which is typically limited to visualization of one layer of the plan at a time. These 2D displays can become difficult to view when plans are large and contain many levels. This paper describes a method of visualizing plans in a three-dimensional (3D) display, allowing both macrolevel understanding of a plan and a detailed view of selected pieces. Constraints within the plan are shown as connections between tasks. The 3D display allows the plan to be viewed from multiple perspectives (sequential tasks, parallel tasks, and abstraction/decomposition relationships)