Tradeoffs in making the behavior-based robotic systems goal-directed

A number of behavior-based autonomous robots with varying degree of reactive functionality have been built. The principle of avoiding an explicit representation of goals inside these robots has limited their achievements to navigation. As a result reactive functionality and goal fulfilling capability have remained orthogonal. The current autonomous robot architectures focus extensively on modifying the internal computational mechanisms of robots, ignoring the important issue of the tradeoffs in robot-environment-goal fit. We provide results on the latter issue. We develop a model of behaviors, environmental modification and goals. Defining the notion of coupling that captures dependency within the internal structure of a behavior space, we show that more complex goals demand higher coupling or more behaviors or a modification to the environment. We show that making behavior-based robots goal-directed without introducing a planner and explicitly represented goals can also mean making the environment goal-directed or decreasing the modularity of the behavior space by increasing the coupling. We argue that these tradeoffs show a spectrum of architectures for integrating goals and behavior-based functionality.