Methodologic Assumptions in ATLANTIS

Gat integrates some of the reactive aspects of the subsumption architecture of Brooks with some of the deliberative aspects of other higher-level models in order to produce a robot capable of collision-free navigation through dynamic environments in pursuit of explicitly represented high-level goals.

Gat is concerned with time and resource management within the robot. He argues that activities can be classified into a hierarchy which displays positive correlation among time commitment, computational complexity and abstraction. The highest level activities are the most abstract, require the most computation and, upon commitment, require more of the robot's time than decisions made at a lower level. Gat motivates this notion with the example of driving a car where the deliberative decision of where to go is more comupationally expensive, more abstract and more time-committing than are the just-as-important but more reactive second-to-second decisions made while controlling the vehicle.

Gat uses these ideas to propose that a goal-oriented, reactive robot can be built upon an architecture that separates actions into low-level and high-level, thereby facilitating both navigation (using reactive mechanisms) and planning (using deliberative mechanisms).


Return to the Top of this architecture.