Planning by transformational synthesis

The results of applying transformational synthesis to planning missions and routes for multiple robotic vehicles are described. Transformational synthesis is an integration methodology that allows many control flow decisions to be deferred until later in the system life cycle, thus enhancing the flexibility and extensibility that one is trying to achieve by using planning technology. The planner has the domain knowledge and planning heuristics to plan multivehicle reconnaissance missions for DARPA's autonomous land vehicle. Most knowledge about the environment and all knowledge about the effects of vehicle operations is uncertain and is represented in terms of probability distributions. Implementation experience with the planner is described.<
>